Oakland is a data consultancy with a 40-year legacy rooted in rigorous research and operational excellence. We wanted to investigate the UK’s readiness for AI, so we embarked on this study with YouGov’s help to help us cut through the hype. 

Research has always been a cornerstone of keeping up with modern business. We centre everything we do around our customers rather than relying on the perspectives of technology vendors and analysts. The report focuses squarely on businesses navigating the realities of integrating generative AI into their operations. Here’s what we discovered – and why it matters.

Generative AI The Hype vs. Reality Gap

A disconnect between potential and practical application marks the generative AI landscape. Businesses are awash with promises of revolutionary capabilities, yet many lack the foundational skills, processes, and AI governance frameworks necessary to make those promises a reality. Our report supported what we’ve found on the ground from speaking to many organisations: most companies aren’t yet in a position to really take advantage of the technology. 

This isn’t something we are surprised by, and it is, in fact, something we’ve seen before; buzz around new technologies often masks the complexity of real-world implementation. Businesses must navigate the technical challenges and the organisational readiness to adapt.

This mismatch is particularly evident in the UK, where many still grapple with fundamental data challenges in their everyday jobs. From data quality issues to incomplete data pipelines to insufficient expertise, the prerequisites for successful generative AI adoption remain unmet for many.

Generative AI’s Two Waves of Innovation

Oakland’s findings align with a broader observation about technological revolutions: they unfold in two distinct waves. The first wave sees the emergence of the technology itself – in this case, large language models (LLMs) and generative AI tools like ChatGPT and Microsoft Copilot. The second wave, however, is where real transformation occurs. This is when we work out what to actually do with this technology and develop meaningful applications that integrate the technology into processes and products. It’s a phase that demands more than excitement; it requires thoughtful engineering, governance, cultural adaptation, and realistic expectations.

The impressive capabilities of LLMs are currently best demonstrated in scenarios like chatbots or simple content generation. However, translating those capabilities into reliable, high-value enterprise solutions remains a significant challenge. Showcasing that we have a really powerful intelligence is a very different challenge to embedding it into business processes especially ones that are of sufficient importance to the business to drive a return on investment that everyone is happy with. 

Our AI consultancy generates value from your data to help you work smarter and deliver meaningful business transformation. 

Find out more here.

Hard Lessons from Early Generative AI Adoption

The report highlights a cautious approach among practitioners, many of whom have scars from previous waves of technological over-promise. Data science and machine learning – two fields that experienced similar hype cycles – serve as cautionary tales. Data leaders have had their fingers burnt before, so they understand the pitfalls of inflated expectations and the high costs of overinvestment in unproven solutions.

Oakland’s findings show that early adopters in the UK are deliberately narrowing their focus to manageable, well-defined use cases. For example, generative AI excels in tasks like reading product descriptions to categorise items or identifying abbreviations in catalogues. These tasks are repetitive and internally focused and benefit from generative AI’s ability to handle high volumes of structured input. However, these projects require significant expertise and investment to deliver tangible results.

The Cost of Complexity

Despite AI’s potential, generative AI solutions are not plug-and-play. Custom applications require weeks of engineering effort, rigorous testing, and extensive oversight. 

“Even a seemingly straightforward proof of concept can quickly escalate into a six-figure project, which, in today’s climate, where budgets for R&D and new technology are often the first activities put on hold, can cripple activity. For small and medium enterprises, this level of investment can be prohibitive, underscoring the divide between early adopters and those waiting on the sidelines to see what happens.” 

Joe Horgan – Oakland Principal Consultant.

Learning from the Plateau of Productivity

For businesses taking a “wait and see” approach, there is wisdom in watching early adopters navigate these challenges. This mirrors other technological journeys, such as electric vehicles, where initial teething problems gave way to broader adoption as solutions matured and costs decreased. Similarly, the report predicts that generative AI’s most exciting phase will emerge post-hype – once expectations are tempered and businesses focus on realistic, high-value use cases.

So, what steps should a business take to become ready for generative AI implementation?

Building a solid business case for Generative AI is fundamental:

1. Be Strategic with Use Cases: Focus on simple, repetitive tasks where generative AI can deliver immediate value. Avoid overly ambitious projects that hinge on unproven capabilities. For example, Generative AI is brilliant at performing simple tasks that would take humans far too long to do. 
2. Invest in Expertise: Whether through hiring or developing partnerships with people like Oakland, having skilled AI engineers is critical to navigating this complex landscape. The pace of change with this technology is mind-blowing, and you need to be able to keep up.
3. Set Realistic Expectations: Understand what generative AI can and can’t do today. The technology’s future potential is vast, but meaningful application requires a grounded approach. You need to know what the LLM is capable of before you can assess if it can solve your problem.
4. Learn from Others: Monitor the successes and failures of early adopters. Use their insights (mistakes) to inform your strategy and reduce the risks of premature investment.

But what exactly are the benefits of using Generative AI and Intelligent Agents in your business? Read our blog to find out!

Is Your Data Ready for Generative AI?

Generative AI represents a powerful new tool in the digital toolbox, but it is just another tool. The massive hype is quite unhelpful because it creates ridiculous expectations. People throw money at it, and it creates bad vibes by diverting resources into older, more proven technologies. It’s when the technology appears on Gartner’s trough of disillusionment that you can start to think clearly and seriously about your Generative AI initiatives and how they can help drive your operational efficiency and competitive advantage. Its transformative potential will only be realised through careful, deliberate application. As the UK moves beyond the initial frenzy of excitement, organisations have an opportunity to define the second wave of innovation – where the focus shifts from possibility to productivity.

Oakland’s Plugged In or Left Out report underscores the importance of patience, planning, and pragmatism in navigating this transformative era. The report highlights that while generative AI’s capabilities are impressive, they are often limited in scope today. For example, simple, repetitive tasks like labelling products or categorising data represent the “low-hanging fruit” where AI can currently excel. However, even these cases require a disciplined approach to implementation.

The road ahead may be longer than the hype would indicate, but it also offers greater opportunities for sustainable growth. By focusing on realistic use cases, investing in expertise, and learning from early adopters, businesses can position themselves in a prime position to get the sort of meaningful value that the businesses’ key stakeholders will be satisfied with. For those willing to embrace these principles, the future of generative AI is bright.

The post Plugged In or Left Out: The UK’s Readiness for Generative AI appeared first on Oakland.

As a Microsoft Data & AI partner, we use Microsoft products daily, with Azure as our preferred technology stack. Microsoft Copilot Studio offers numerous benefits, particularly its seamless integration with Microsoft 365 apps and the utilisation of large language models (LLMs) like GPT-4. If you are looking to introduce artificial intelligence into your organisation across various use cases, Copilot Studio is an excellent starting point for AI integration.

If you want to learn more about Oakland’s Intelligent Agents, please click here to read our blog.

What Are the Key Features and Benefits of Microsoft Copilot Studio?

Microsoft Copilot Studio excels in several areas, making it a strong candidate for many organisations starting their journey with Generative AI. Copilot Studio is capable of addressing a wide range of business needs:

Functionality

• Knowledge Source: Through Microsoft Graph connectors, Copilot Studio connects to a wide range of data sources, both internal (SharePoint, Dataverse, Fabric) and external (public websites, Confluence, Jira, Excel, databases).
• Real-Time Data Access: Ensures up-to-date responses by retrieving real-time data from integrated sources.
• AI Integration: Enhances its functionalities by integrating with Azure OpenAI models.

Workflow Automation

• Automated Workflows: Users can create and manage topics that trigger based on specific inputs or events.
• Actions: Copilot Studio can perform tasks such as updating SharePoint lists based on user inputs, and streamlining operations.
• Integration: Integrates effectively with other Power Platform tools, expanding its capabilities.

User Management and Security

• Role-Based Access Control: Defines user roles and permissions to ensure secure access.
• Audit Logs: Maintains detailed logs of user interactions and system activities, ensuring transparency and security.
• Single Sign-On (SSO): Quick setup of SSO for accessing relevant data sources, enhancing user experience and security.

Deployment

• Channels: Copilot Studio can be deployed across various platforms such as websites, Teams, and Slack, making it highly versatile.
• Handoff: It can hand off chats to third-party chatbots or other Copilot instances like Zendesk or Salesforce, ensuring seamless user experiences.

Reporting

• Power BI Reports: Built-in reports showcase Copilot Studio’s performance and effectiveness, helping organisations track and optimise usage.

What Are the Limitations of Microsoft Copilot Studio?

Despite its strengths, Microsoft Copilot Studio has some limitations that may pose challenging for certain organisations:

Setup Complexity

• Configuration Overhead: Setting up data sources and configuring Copilot Studio can be complex and time-consuming, especially for users with limited technical expertise.
• Integration Challenges: Integrating with legacy systems or custom applications may require additional development and troubleshooting, adding to the setup complexity.

Natural Language Understanding Limitations

• Ambiguity Handling: Copilot Studio may struggle with ambiguous or poorly structured queries, leading to incorrect or incomplete responses.
• Language Support: While Copilot Studio supports multiple languages, the accuracy and depth of understanding can vary between them, potentially limiting its effectiveness.
• Incorrect Topics: Occasionally, Copilot Studio elects the wrong topic, resulting in inconsistent answers.

Performance and Scalability

• Response Time: Copilot Studios response time can be affected by the performance of integrated data sources and the complexity of workflows, which may not be ideal for high-demand environments.
• Scalability Limits: Some sources use GraphSearch instead of VectorSearch, leading to timeouts as data points scale up, which could hinder performance as your business grows.

What Are the Use Cases for Microsoft Copilot Studio?

Here are some practical use cases where Microsoft Copilot Studio can be effectively deployed:

Microsoft Teams Bot

Create a Microsoft bot with SSO enabled and internal knowledge attached to the Copilot. For example, an HR Copilot with information on HR-related topics like sick leave and holidays.

Power Platform Users

Teams already using Power Platform apps like Power Apps and Power Automate can easily create Copilots using Copilot Studio, leveraging existing tools for enhanced functionality.

Action Bot

Copilots can perform actions like updating SharePoint lists or sending emails, making them useful for creating GenAI bots with built-in actions.

Simple Website Bot

Attach a Copilot to a company website for a bot powered by Copilot and GenAI, providing users with instant, AI-driven support.

“Oakland are big fans of the Microsoft ecosystem, and Copilot Studio has proven to be an excellent platform for exploring GenAI capabilities. It allows us to create multiple co-pilots tailored to different client needs, and we’ve been impressed by its seamless integration for straightforward use cases. However, when it comes to more complex solutions requiring advanced capabilities, Oakland’s Intelligent Agents excel in delivering value quickly and accurately.”

Rehan Hussain – Senior Solutions Architect

Custom-Built Intelligent Agents: The Next Level of AI Integration

While Microsoft Copilot Studio offers a robust starting point, custom-built intelligent agents provide several distinct advantages tailored to your business needs. Here are some key advantages:

Customisation and Flexibility

• Custom-built agents are designed for a business’s unique needs and workflows, ensuring perfect alignment with business processes and objectives.
• Businesses can define precise functionalities and features critical to their operations, which Copilot Studio may not fully cover.
• Custom agents can be easily modified and updated to accommodate changing requirements, new business processes, or emerging technologies, ensuring long-term effectiveness.

Data Integration and Control

• Custom agents integrate deeply with various internal systems, databases, and third-party applications beyond the Microsoft ecosystem, providing a more comprehensive solution.
• Organisations maintain full control over their data, ensuring compliance with specific regulatory requirements and internal policies, which is crucial in highly regulated industries.
• Custom-built solutions are designed with specific security protocols and measures tailored to an organisation’s risk profile, offering superior protection to generic solutions.

Optimised Performance

• Custom agents are optimised for the specific types of tasks and queries most relevant to the business, leading to potentially faster and more accurate responses, which is critical for mission-critical operations.
• Businesses can allocate computing resources and manage infrastructure to ensure the optimal performance of their intelligent agents, maximising efficiency and effectiveness.

Cost Efficiency

• While initial development costs may be higher, custom solutions can be more cost-effective in the long run, avoiding ongoing pricing increases and subscription fees associated with services like Microsoft Copilot Studio.
• Businesses can invest incrementally in features and capabilities based on their budget and priorities, allowing for a more controlled and scalable investment in AI technology.

Branding and User Experience

• Custom intelligent agents can be branded and designed to reflect the company’s identity, providing a consistent user experience aligned with business values and aesthetics.
• The user interface can be tailored to fit the specific needs and preferences of the business’s workforce, enhancing usability and adoption, which is key to ensuring widespread use and acceptance within the organisation.

Innovation and Competitive Edge

• Custom-built agents can incorporate proprietary algorithms, models, and technologies that provide a competitive edge not available to competitors using standard solutions.
• Businesses can innovate and develop unique capabilities that set them apart in their industry, creating opportunities for differentiation and market leadership.

The Oakland Advantage

Microsoft Copilot Studio offers a powerful and integrated AI solution within the Microsoft 365 ecosystem, making it a suitable choice for many organisations. 

However, a custom-built intelligent agent provides unparalleled customisation, data control, performance optimisation, and potential cost efficiencies. This can make it a compelling choice for businesses with specific needs, high-security requirements, or unique workflows that are not fully addressed by off-the-shelf products like Microsoft Copilot Studio.

At Oakland, we are at the leading edge of building intelligent agents that can be deployed in weeks rather than months. Our cost-effective solutions are transforming businesses, providing tailored solutions that drive innovation and competitive advantage.If you’d like to hear more about Oakland’s Intelligent Agent Eco-system, please email hello@oakland.com or book a meeting with one of our AI experts here.

The post Advantages of a Custom-Built Intelligent Agent Over Microsoft Copilot Studio appeared first on Oakland.

Yet, for enterprises to experience the benefits, finding the right use cases is critical. We have extensive experience delivering custom AI solutions for some of the UK’s leading businesses. This means our Data and AI consultants have developed a deep understanding of when generative AI should and shouldn’t be used, particularly in knowledge management.

Below, we explore how generative AI can help enterprises scale the knowledge management mountain, overcome the tidal waves of data modern businesses are lashed by, and tap into its significant value.

What is AI in Knowledge Management?

Knowledge management – the process of gathering, organising, and distributing information within an organisation – incorporates employee knowledge, processes, records, feedback, product information, and so much more. When generative AI is applied to knowledge management systems and processes, significant portions are essentially automated.

The result is knowledge management that automatically catalogues and categorises information. Users can access information far quicker and easier. With every piece of data added, query submitted, and piece of information served, the system learns. It becomes smarter, faster, and more efficient. It ultimately becomes tailored to the specific nature of the business using it, driving knowledge retention, employee collaboration, customer service, and performance.

According to IBM research, 38% of enterprise-scale companies (those with over 1,000 employees) reported using generative AI in their business in 2024, and another 42% were exploring the technology. Knowledge management was stated as a centre of attention, but with competition for better practices ramping up, what use cases can companies new to the technology focus on?

Key Generative AI Knowledge Management Use Cases & Benefits: Document Mountain And Information Tsunami

Generative AI has emerged as a game-changing tool for addressing two of the biggest challenges in modern knowledge management: the Document Mountain and the Information Tsunami. These two problems, while related to the overwhelming amount of data businesses generate, present distinct challenges that generative AI is uniquely suited to solve.

What is the Document Mountain Knowledge Management Problem?

One of the most prevalent and complex challenges we’ve encountered is what we call the Document Mountain – vast repositories of knowledge that lack management and are seemingly unmanageable.  Solving this challenge with generative AI can unlock the value held in data such as ‘lessons learned’ documents to proactively generate insights that can hone your business’s operations and prove genuinely transformative.

In many large enterprises, effective knowledge management is a critical yet elusive goal. It’s a classic “big problem” that affects many business areas but is notoriously difficult to solve. 

Health and safety, risk and compliance, procurement, bid management, quality control, and project management: all key functions , but all heavily reliant on the efficient sharing of knowledge. Without it, mistakes are repeated, risks go unnoticed, and opportunities slip away.

These failures in knowledge management are all too common in large organisations. Learning loops are often broken, leading to ineffective knowledge transfer and, ultimately, costly errors. But why is this the case? 

For most businesses, the challenge isn’t a lack of documentation. On the contrary, many organisations have invested significant resources in creating detailed records of their observations, learnings, and processes. The problem lies in the overwhelming volume of documentation they now generate – the Document Mountain.

Examples of these documents include:

• Safety incident reports
• Internal or external policies and guidelines
• Project review “lessons learned” documentation
• Customer correspondence and complaints
• Invitations to Tender (ITTs), Requests for Proposals (RfPs), bid responses, and feedback
• Non-compliance reports
• Project documentation, such as status reports
• Maintenance manuals and reports.

These documents are often stored in vast, fragmented libraries that are difficult to navigate and access. The insights users need are in there, but finding them within the available time can seem almost impossible. And to make matters worse, in some cases, it’s a mountain range, as documentation is scattered across different repositories and storage systems. 

The result? 82% of organisations report their data is siloed and 24% don’t trust it. Their employees then spend, on average, 2 hours a day searching for the information they need.

What makes this problem particularly challenging is the nature of the documentation itself. Much of it is unstructured free text, which traditional search methods or analytical techniques struggle with if documents lack metadata.

As a result, the valuable insights contained within these documents remain trapped and inaccessible to those who need them. 

The scale of this problem is truly staggering. According to International Data Corporation analysis, 80% of the world’s data (140 zettabytes in all) will be unstructured by 2025. 

The impact of these broken learning loops can be just as enormous. Project overruns, safety incidents, regulatory failures, and lost business opportunities – all of which can have significant reputational and financial consequences. A recent analysis cited by Fast Company found that, in the US, Fortune 500 companies lose around $31.5 billion each year from the effects of their knowledge siloes.

Generative AI: the Solution to the Document Mountain

The powerful natural language processing capabilities of generative AI and large language models allow enterprises to unlock the insights hidden within their Document Mountains.

Generative AI excels at reading, interpreting, and summarising large volumes of free text and can support easy, natural-language interactions with users. 

Imagine being able to ask questions of your vast repositories of unstructured data and having the AI respond with relevant, actionable insights. What would that data reveal if it could talk back?

Scaling the Knowledge Management Mountain: Key Use Cases

Let’s explore some specific use cases where generative AI can help enterprises solve the problem for good.

1. Customer Service

Delivering exceptional customer service requires having the right information at your fingertips. This can be particularly challenging in sectors like banking or insurance, where customers expect quick answers to complex questions about their terms and conditions or coverage details.

Generative AI can create virtual assistants that assist customer service agents in real time, providing them with accurate and relevant information from vast stores of documents. By augmenting customer-facing teams with AI-powered tools, organisations can significantly improve response times and service quality.

2. Complaint Management

Handling customer complaints and feedback is another area where Generative AI can shine. 

Complaints and feedback often arrive as unstructured data at high velocity, making them difficult to manage and analyse effectively. They also often need to be read and compared with complex, nuanced policies and regulations. As such, human teams are overwhelmed by the sheer volume of work, and simple, rules-based complaint-handling software often struggles with the complexity of the task.

Generative AI is well-suited to reading, summarising, and categorising this information quickly. This capability allows organisations to handle new complaints at scale while also extracting valuable insights from the broader dataset. By equipping generative AI with relevant policies and operating procedures, companies can even build AI agents that not only process complaints but also respond appropriately.

3. Bid Management and Bid Development

In industries where companies bid on major contracts, vast libraries of bid responses, case studies, credentials, service specifications, and similar documents often accumulate over time. 

Effectively utilising this information remains a long-standing challenge. Best practices are only sometimes shared, and content is frequently duplicated or recreated from scratch. Additionally, bid assessment feedback is often under-analysed, leading to missed opportunities for improving future responses or service design.

Generative AI can assist by analysing and summarising previous submissions, highlighting best-practice examples for reuse, and speeding up the bid development process. By improving the quality of tender responses, AI can help organisations win more business while reducing the time and effort required to prepare bids. 

In large organisations competing for multi-million-pound tenders, even small gains in bid management effectiveness can reap huge ROI on the investment in Gen AI solutions.

4. Lessons Learned

Effective knowledge management is a classic challenge for many enterprises, regardless of whether they have a dedicated Knowledge Management team. Businesses often accumulate knowledge in the form of “lessons learned” documents, project updates, incident reports, guidelines, policies, and more. However, the sheer volume makes it impossible for individuals to read and consume all the available information.

Generative AI can unlock the insights within these documents, allowing users to query them through a natural language interface. By providing quality-controlled responses, AI ensures that only valid and relevant lessons are shared, helping organisations to learn from past experiences and avoid repeating mistakes. Learn how we provided this for Network Rail.

It’s partly for this reason that, in IDC’s 2022 Knowledge Management Strategies Survey, improved business execution was the top benefit experienced by businesses that had implemented knowledge management systems, driving significant AI return on investment.

5. Health and Safety

Health and safety is a critical area in which businesses must navigate complex rules and policies. Front-line operatives and managers often need help accessing relevant information when they need it most. Generative AI can be used to create a virtual “Safety Assistant” that makes health and safety information more accessible and actionable.

Additionally, compliance monitoring is another area where AI can make a significant impact. By scanning project documents, job reports, incident logs, and other relevant materials, generative AI can identify early signs of non-compliance or potential risks, allowing H&S teams to address issues before they escalate into serious incidents.

There are countless use cases for generative AI, but it’s crucial you tailor yours to your challenges. Learn more about how to find your specific use cases.

Read more

What is the Information Tsunami Knowledge Management Problem?

Many large businesses struggle with knowledge management, and it’s a significant challenge that can disrupt critical processes if not addressed effectively. Traditionally, when we think of knowledge management, we often focus on the volume of information – the Document Mountain detailed above. But the issue goes beyond just managing large volumes of data. The speed at which it comes in can create problems of its own: this is the Information Tsunami.

In a world where speed is essential, information doesn’t just trickle in slowly like pages in a library. It can come in fast and furious, and without the right tools, it becomes overwhelming. This is where many businesses get stuck. So, what is the Information Tsunami, and how do we solve it?

The Cause: Real-Time Monitoring Systems

Many businesses today have invested heavily in real-time monitoring systems to keep track of their critical processes and operations. These systems are made possible by more affordable software, sensors, and IoT technologies. Examples include:

• Telemetry systems for physical networks (pipes, vehicles, or assets)
• Environmental sensors (temperature, traffic, or footfall)
• Live customer chat and feedback monitoring
• Real-time supply chain and product management systems (inventory, product quality, manufacturing processes)
• Smart products using IoT technology that send real-time performance data and alerts.

These monitoring systems create high volumes of data at high velocity. This information needs a fast response if it’s to be used effectively. Unlike ‘Document Mountain’, this is about responding quickly and decisively. In front-line, fast-paced operational or commercial environments, simply gathering the information and reflecting on it later is not an option.

By collecting real-time data, businesses can prevent problems or address them quickly. This sounds great in theory, but in practice, it often leads to information overload.

The Problem: A Flood of Data

What usually happens is that control teams are quickly overwhelmed by a flood of incoming information from these monitoring systems. Imagine being bombarded by constant alerts – many of which are minor or irrelevant – while trying to spot the critical ones that need immediate attention. Necessary signals get lost in the noise.

The result is often a failure to act on vital information, leading to significant operational or financial costs. Businesses that have invested heavily in monitoring systems may not see the return on investment they were hoping for because they can’t effectively manage the sheer volume and speed of incoming data. It’s like trying to find a needle in a haystack while the hay keeps piling up around you.

Previous Solutions and Their Limitations

In response, many companies have tried to solve the Information Tsunami using various technologies. Some common approaches we come across include:

• Rules-based prioritisation of incoming alerts
• Automated event-handling systems
• Anomaly detection techniques
• Machine learning models designed to filter out the noise.

While these solutions work in simple scenarios, they often fall short in more complex environments. Here’s why:

1. Rules-based systems struggle with the complexity of handling real-world incidents, which often requires more nuance and flexibility than simple if-then rules can provide.
2. Event-handling policies are frequently stored as free text documents, making them difficult to interpret by statistical or machine learning models or translate into simple ‘if-then’ rules.
3. These systems often can’t dynamically respond to specific user queries or instructions or be tailored to specific users, limiting their usefulness in fast-paced control and customer service environments.
4. Effective alert management requires context from secondary data sources like free text maintenance reports and querying supporting data stores, many of which are unstructured. Existing systems can’t easily retrieve or analyse this data.

Because of these limitations, these solutions often generate too many exceptions or miss important alerts. They’re complicated and time-consuming to use. Or generate incorrect actions. As a result, control teams are still overwhelmed, and trust in these systems erodes.

Taming the Information Tsunami with Generative AI

Generative AI offers a new approach to solving the Information Tsunami. While generative AI is often associated with processing large amounts of static data, such as the Document Mountain, its ability to handle real-time high-velocity data makes it uniquely suited to managing the Information Tsunami.

The natural language processing capabilities of Large Language Models (LLMs) allow them to respond dynamically to complex, real-time data streams. When combined with machine learning and other analytical techniques, LLMs can interpret data in a more human-like way. This flexibility allows AI systems to interact with users through natural language, pulling in information from various structured or unstructured sources to provide richer, more relevant responses.

This makes generative AI a much more powerful tool for handling the Information Tsunami, allowing human teams to focus on critical issues that require immediate attention.

Use Cases for Generative AI in Managing the Information Tsunami

Let’s explore a few real-world examples of how generative AI can be used to manage the Information Tsunami:

1. Quality Management in Supply Chains

Even minor quality issues in a supply chain can cause significant disruptions if not addressed quickly. Generative AI can help by monitoring these issues in real time, assessing their downstream impacts, and proactively alerting the appropriate teams. The AI can analyse complex product documentation and organisational structures to ensure the right information reaches the right people before the problem escalates.

2. Operational Control

In operational environments, AI can act as an alert triage manager. It can filter incoming alerts based on protocols, automatically handling minor issues and escalating more critical ones to human operatives. This ensures that teams aren’t overwhelmed by unnecessary data and can focus on high-priority cases. Additionally, the AI can advise on proper response procedures by referencing relevant policies or regulations.

3. Data Governance

In modern enterprises, Agile and Dev Ops practices mean data and systems are constantly evolving. As a result, tracking changes and maintaining all of these is so time-consuming as to be nearly impossible. While automated data quality monitoring has existed for a while, generative AI takes it further by updating critical data governance documentation, such as glossaries, lineage maps, and catalogues, in real time. 

This automation augments human teams relieving them from time-consuming tasks. It ensures that documentation remains current, even as data systems evolve, helping flag non-compliance and errors – a truly powerful capability.

4. Technical Oversight

Maintaining compliance with security and architectural standards is a massive challenge in organisations with hundreds of software and data engineering teams. AI can assist by reviewing design submissions for compliance and flagging exceptions. Over time, this increases adherence to standards without requiring significant resources from central oversight teams.

See how we revolutionised knowledge management for Network Rail with generative AI.

Hone Your Knowledge Management Strategy with Intelligent Agents

Generative AI holds enormous potential to tackle both the Document Mountain and the Information Tsunami. When implemented effectively, it can transform knowledge management, streamline operations, and unlock the value hidden in your data.

While generative AI is a powerful tool, we believe that the best way to solve these complex use cases is through the deployment of intelligent agents. These AI-driven agents are designed to tackle specific challenges within the enterprise, combining the power of generative AI with other advanced technologies to deliver even greater value.

Check out our guide for more insights into why we recommend intelligent agents as a solution to these kinds of problems.

Make Your Data Pull Its Weight With Oakland’s Generative AI Experts

Document Mountain is a prime example of a problem that generative AI is uniquely suited to solve. By leveraging AI to unlock the insights trapped in vast repositories of unstructured data, businesses can drive operational efficiencies, enhance decision-making, and ultimately achieve better outcomes.

In future blogs, we’ll explore other common enterprise use cases for generative AI. But for now, remember: the key to success with AI lies in choosing the right problems to solve. And when it comes to scaling the Document Mountain, generative AI is the solution you’ve been waiting for.

Learn more about our approach to AI, then contact our experts to learn how their extensive experience working with enterprises can be leveraged to benefit your business. 

The post How to Improve Your Knowledge Management Strategy with Gen AI appeared first on Oakland.

Are you yet to embrace the benefits of AI in your enterprise? Now is the time, and one of the the most effective forms of AI to implement into your business is Generative AI. Generative AI is responsible for some of 2024’s most well-known AI agents. ChatGPT and Gemini, to name just two, which have certainly taken the world by storm. 

These large language learning models specialise in understanding and generating natural language, making them ideal for powering chatbots in business settings. However a chatbot is far from the only way to use this brilliant technology, though. In this guide, we’ll explore all the clever ways you can make Generative AI your business’s number one secret weapon.

What is Generative AI (and why all the hype?)

In simple terms, Generative AI is a form of artificial intelligence capable of generating text, images, videos, or audio. The form of Gen AI that has grabbed the most headlines is the text-generating large language model (LLM). Examples of LLMs you may have heard of include Chat GPT (Open AI), Gemini, LLaMA, and Mistral. 

Large Language Models are hugely powerful deep-learning models. Their vast training sets give LLMs natural language processing abilities that exceed previous AI technologies.

What is NLP?

Natural Language Processing (NLP) is an AI technique that focuses on enabling computers to understand, interpret, and generate human language. NLP combines linguistics, computer science, and machine learning so it can process and analyse large amounts of natural language data.

Because they understand natural language so well, LLMs grasp and act on free text instructions, you can chat with them and direct them like a human. This slashes development times and avoids the extensive programming for previous AI technologies. It’s this leap in productivity and ease of execution that’s so exciting for enterprises. 

In short, combining AI and natural language processing makes artificial intelligence accessible to non-technical users and much cheaper to deploy.

What Can Businesses Use Gen AI for?

At Oakland, we are a data consultancy with a wealth of experience in everything data; from strategy and governance to data  engineering, analytics and artificial intelligence. One thing we feel incredibly passionate about is simplifying data strategy by taking an easy, seamless approach that everyone in your organisation can understand. 

What Not to Do

Imagine the huge amount of insight trapped within your business stored in free-form texts, videos and images. Now you can draw insights from these ‘hidden’ datasets. This vast data landscape has long been a double-edged sword for businesses—rich in information but daunting in scale and cost to analyse. Critical insights or risks are missed because they are buried in the data. AI represents a revolutionary shift in data handling because it can read, summarise and generate content at a scale unimaginable by human workforce.

Businesses can now harness the advantages of Generative Artificial Intelligence to build their own solutions and applications. The AI services offered by Open AI, Azure Cognitive Services, AWS, and others allow enterprise users to call on large language models via API.

The natural language processing power of LLMs makes them well suited to a range of tasks, including:

• Interrogating data or public information sources to extract insights and summaries
• Summarising or extracting meaning from large volumes of unstructured information, such as free text
• Writing or checking code (developer assistance)
• Recognising patterns or objects in texts and images
• Executing rule based or repetitive tasks
• Validating documentation against standards
• Chatting in structured contexts (e.g. Q&A, user or customer support)

How Could We Implement Generative AI At Our Enterprise?

Enterprises looking to adopt Generative AI already face a range of technology options. But if you strip it back, there are three basic choices:

• Public models (e.g., Chat GPT online): allow you to simply access a website over the internet and chat with an LLM via your web browser. For example, the Chat GPT website or Co-Pilot online. However, these interactions are not private and typically not suitable for enterprise users.
• Plug-in tools: provide pre-packaged Generative AI capabilities. Some are highly specialised solutions for specific industries or use cases, but others aim for a wider enterprise user base. Perhaps the most famous is the Microsoft 365 Co-Pilot. 365 Co-Pilot provides Generative AI functionality via desktop Microsoft applications such as Teams, Outlook, and Word.
• Custom Generative AI Solutions: many enterprise use cases can’t be solved with public or plug-in AI tools. For most businesses, building your own LLM is the stuff of dreams. It’s too expensive and time-consuming. Thankfully, there is now a range of services that allow you to build your enterprise Generative AI applications. We could blow your mind with all of the details, but these services allow you to harness Large Language Models to build solutions and embed Generative AI into workflows. How much customisation and configuration you do is determined by your resources and chosen use case.

Always Start With The Use Case!

The Gen AI landscape is full of skeletons of proof of concepts that didn’t stick. If you want to make a difference, start with a clear use case. What is the clear and specific problem that you want to solve?

Finding the right use cases is key. But you need to know what to look for.

Every business is different. We’ve suggested some examples for different business functions further down this article. There are also some common characteristics of use cases for Generative AI that you can look out for. 

In our experience, there are three key problem areas that Generative AI applications are great at solving:

• Unlocking your document mountain: the same mistakes occur time and again in businesses because knowledge isn’t shared. Often, hours are spent gathering documentation about ‘lessons learned’, best practices or incident reports. But the information is just too big and messy for anyone to use it. Important lessons are lost in the noise. If you can unlock Knowledge Management with Generative AI, you can massively improve process execution, compliance and knowledge sharing. 
• Information tsunami: many businesses have invested in monitoring systems for their systems or assets, but now face an overwhelming volume of alerts and alarms. Monitoring and control teams are endlessly distracted and can’t separate the genuine alerts from the noise. Generative AI is great at scanning alarms and alerts and, crucially, relating them to operating procedures and policies so that informed decisions can be made.
• Expert shortage: data is so often meaningless without context. But, applying that context or audience-specific messaging is time-consuming. How many warning signs are lost in the data or missed because nobody spots their relevance? Many companies have implemented human ‘business partnering’ models in functions like Finance, HR and IT to connect end users to expertise, but these models are expensive and often overstretched. Generative AI is great at preparing reports, highlighting insights and answering detailed user questions in a relevant way. It has great capabilities in analyst and business partnering contexts.

Using an Intelligent Agent in AI

The buzz around generative AI is undeniable – but so many of the organisations we speak to are still struggling to find the right use cases, and prove the ROI which could be why Gartner found that nearly half of AI projects fail to meet their goals. 

Finding a great enterprise use case for Gen AI is a start. But you also need the right solution. Large Language Models provide amazing capabilities, but you have to harness them in the right way. Chatbots are a great example of an LLM-powered solution, but they won’t solve every problem. If you want to really unlock Gen AI adoption at your enterprise, you need solutions that can do more.

At Oakland, we build Generative AI solutions with Intelligent Agents. Intelligent agents are AI-powered programs powered by large language models, but they also have powerful tools, careful task orientation, and context awareness. The easiest way to think of them is as highly scalable virtual workers. You can read more about Intelligent Agents and how we use them to solve complex business problems in our expert guide. 

With Intelligent Agents, you stand a good chance of solving some of the use cases spelt out below.

Specific Generative AI Use Cases

Nothing beats a concrete example, so let’s explore some specific enterprise use cases for Generative AI. These are organised by business function:

Customer Experience

Customer chatbots are a classic use case for artificial intelligence and natural language processing. Generative AI can elevate these to the next level because of the ease and conversational nature of the interactions they can support. And the good news? chatbots based on LLMs require far less training and testing than previous generations of natural language processing techniques. 

Complaint or feedback handling is another exciting use case. Complaints or customer feedback often means high-velocity unstructured data that could come in at any time. Generative AI is great at reading, summarising and categorising this information. This allows you to handle new complaints at speed and scale, but also extract detailed insights from the wider dataset of all complaints.  

Equipping generative AI with policies, operating procedures, and communication capabilities allows you to build a ‘Complaint Manager’ intelligent agent that not only reads and categorises incoming information but also acts on it.

Knowledge Management

Unlocking knowledge management is a classic challenge for many enterprises, whether they have a dedicated Knowledge Management team or not. Most businesses have a document mountain piled high with  ‘lessons learned’ documents, project updates, incident reports, guidelines, policies and the list goes on. But nobody can read and consume them – they’re too big. So the knowledge never makes it to where it’s needed.

Generative AI is a brilliant tool for unlocking this kind of information. You can use it to summarise information for users quickly. Users can query complex unstructured datasets through a conversational, natural language interface. 

This unlocks and democratises information in a way that would have been prohibitively time-consuming or expensive before Gen AI. 

A properly created Knowledge Management solution powered by Gen AI is like an always-on, knowledge librarian with an endless memory. If you want to go further, you can even empower the AI to proactively scan for new insights in its data store and alert users when they are found.

Health and Safety

Many businesses have complex H&S rules and policies that are difficult for front-line operatives and managers to navigate. Creating a virtual ‘Safety Assistant’ powered by Generative AI is a great way to make that information relevant and accessible. 

Compliance monitoring is another example of where Gen AI “powered applications” can shine in Health and Safety. Project documents, job reports, incident logs etc, can all contain clues or early flags for non-compliance. Usually, H&S teams are too stretched to read these, but a Gen AI solution could be used to scan these documents and highlight risks or non-compliant activities before they become incidents. 

Product Development and Supply Chain

Many organisations have huge volumes of customer feedback and market insights that they need to summarise quickly to drive product development. Generative AI has great potential to streamline the gathering and summarisation of this insight, putting it in the hands of product teams faster than ever before.

Within Supply Chains, even small quality issues or incidents can disrupt entire supply chains if not resolved quickly. Rapidly assessing the downstream impacts and alerting the right people is critical. But the required information maps often don’t exist and can’t be made quickly. So, communications can’t be directed, and issues spiral out of control. Generative AI capabilities can read and analyse complex product documentation, highlighting probable downstream impacts quickly when new issues are discovered. 

Taking it further, a Gen AI-powered Intelligent Agent could monitor quality issues in real-time, rapidly assessing downstream impacts and proactively alert impacted teams based on its knowledge of the organisational structure of the host business.

Legal

Generative AI’s ability to generate new text content has obvious applications for legal departments. Drafting contracts, policies, or letters are all clear and well-described use cases. Generative AI’s creation of documentation following a repetitive, standard format is a clear strength.

Contract analysis and management is a more interesting use case that many businesses are exploring. Many companies have hundreds, if not thousands, of contracts with suppliers, customers and other third parties. Managing and accessing this huge volume of contracts is hugely challenging for human teams. Answering simple questions such as how many customers are on a given version of a standard contract becomes almost impossible with the resources available. 

Using a Generative AI solution with access to the contract dataset, legal teams could quickly search and analyse the entire contract base at once through a natural language interface. This hugely reduces the time taken to answer queries and frees up human teams to focus on more strategic or advisory tasks.

Asset Management

Many enterprises operate complex networks of different types of physical assets. Effective management of these assets relies on having a clear view of their design, installation, maintenance requirements and historical activity. 

In fault scenarios, in particular, this information needs to be rapidly available. But, in many companies, this data is scattered across different systems, often in unstructured and inconsistent formats. 

Generative AI could be used to build an ‘Asset Historian’ that plugs into asset data stores to build a 360 view of an asset and answer detailed user queries or generate asset reports. 

This hugely increases the speed at which asset information can be assembled, in turn shortening manual effort and response times for Asset Management teams. 

Operational Control Room

Many businesses have invested heavily in alarms and telemetry to monitor their asset bases or systems. However, the volume and velocity of these alerts are so high that control room teams are often overwhelmed. Triage and prioritisation either cannot happen or take so long that valuable response time is wasted. The result is critical alerts should be addressed or handled properly.

Generative AI can be used in an ‘alert triage’ role. 

You could create an AI-powered alarm triage manager by showing an LLM alert handling protocol and giving it the correct prompting. The AI would interpret and filter incoming alerts, proactively making simple fixes and passing higher-priority cases on to control operatives. This frees up human operatives from having to try and consume an overwhelming volume of incoming alerts, allowing them to focus on triaged, high-priority alarms. 

Here, the near-limitless attention span and memory of an AI solution have been harnessed to support the ease and efficiency of human teams.

Finance

We’ve already discussed Gen AI’s strengths in helping users navigate large volumes of complex documentation. These strengths have clear applications within the world of finance. For example, a finance assistant could help non-accountants understand financial rules and policies to prevent errors from happening. Similarly, Generative AI could also scan internal finance policies and guidelines to check for discrepancies or gaps against external accounting policies or regulations. 

Many finance teams also face the crucial but time-consuming task of providing regular financial reporting and commentary, such as monthly performance reviews or other regular financial updates. Generative AI can be leveraged to compile repeatable reporting and add descriptive commentary for users. Employed correctly, a Gen AI solution could also answer follow-up questions on reporting from Enterprise users. This frees expert finance teams to focus on strategic advice and deep insights rather than repetitive drafting.

Generative AI for Data Science

One of the clear strengths of Gen AI models/LLMs such as Chat GPT is their ability to write code in response to user requests.

Many software engineering and data teams are already harnessing this code generating ability – many liken it to having a team of junior engineers to help them. 

Gen AI can rapidly write or review code, implement code changes across different codebases or even convert code from one programming language to another. You can also use LLMs to shorten the time-consuming task of creating and reviewing the technical documentation required to accompany software development.

These are classic examples of ‘AI augmentation’, where Gen AI is used under the supervision of human teams to speed up efficiency and implement repetitive workflows.

These are just some of the use cases of Generative AI there are many many more. The difficulty can be trying to find the early use cases which you can use to prove value quickly. If you need help and guidance on how to find the right use cases or are looking for an AI solution to an known problem then get in touch, and we’ll find the Generative AI solution that’s a perfect fit for you.

The post How to Find Generative AI Use Cases for Enterprise appeared first on Oakland.

At Oakland, we pride ourselves on our extensive expertise in data and our approach to simplifying it. This ensures you can fully understand how your business can reap the benefits of having a strong data strategy.

In this comprehensive guide, we’ll teach you all about the power of data and the right way to approach it to do just that.

What is a Data Strategy?

A data strategy sets a vision and roadmap that explains how an organisation will use data and analytics to achieve its strategic objectives.

A well-defined data strategy aligns data and analytics activity with clear business aligned objectives. This involves outlining the changes, investments, and capabilities needed to make the data strategy effective.

Remember, data strategy is not about creating a different future for your organisation. At Oakland, we want to use data to realise the organisation’s vision and strategy. It’s all about using data to serve organisational goals and drive value.

The wider future of the organisation is (or should be) already imagined through the overall business strategy.

Why Do You Need a Data Strategy?

Simply put, it today’s market, organisations need a data strategy in order to propel them towards their business goals. Here are just a few of the key reasons having a data strategy is essential:

Opportunity Identification

 Many organisations sense missed opportunities with their data but struggle to pinpoint them. A structured data strategy helps uncover and prioritise these opportunities.

Vision Setting

A data strategy creates a clear future vision to unite data efforts and guide decision-making.

Alignment

It aligns priorities, resources, and roadmaps behind a clear vision, addressing disjointed data functions and activities.

ROI Realisation

Ensures investments in data yield expected benefits and helps secure necessary budget and resources.

Customer Satisfaction

Successful data implementation improves data availability, access, quality, and timeliness, addressing common frustrations.

Foundation Building

Links customer frustrations with root causes and provides solutions, fostering data governance.

Breaking Reactive Cycles

Helps data teams move from a reactive stance to a proactive, strategic approach.

If these points resonate with your organisation, it’s likely time to consider creating or updating your data strategy.

Finding the Right Approach to Your Data Strategy

At Oakland, we have an extensive collection of expertise in all things data; from strategy and governance to data  engineering, analytics and artificial intelligence. One thing we feel incredibly passionate about is simplifying data strategy by taking an easy, seamless approach that everyone in your organisation can understand. 

What Not to Do

Bear with us. It sounds strange, but a data strategy shouldn’t be a strategy for data.

It’s so easy to fall into the trap of thinking of data as something separate from the wider organisation, with its own priorities, objectives and concerns. But this is a fatal error.

Take a look at this. Where would you say you are today?

A common mistake is to fall into the trap of creating a ‘strategy for data’ rather than using data to achieve organisational goals. Avoid thinking of data as separate from the wider organisation. Instead, re-imagine data as a means to meet organisational needs.

A Better Way

To avoid the ‘strategy for data’ pitfall, our data strategy consultants work collaboratively alongside your organisation to ensure your data strategy focuses on the following:

1. Organisational Outcomes: Understanding what outcomes your business needs and the role of data in achieving them.
2. Balanced View: Incorporating people, processes, and technology, rather than overemphasising tool and technology alone.
3. Compelling Narrative: Crafting a data strategy that tells a compelling story, making the technical aspects accessible and engaging for non-technical stakeholders.
4. Collaborative Creation: Developing the strategy through workshops, customer engagement, and stakeholder collaboration to ensure alignment and instill confidence.

Oakland’s Approach: What Should a Data Strategy Include?

At Oakland, we have created a foolproof 3-step approach to create a comprehensive data strategy, ensuring it covers the following key components:

1. Strategic Vision

Outline what you aim to achieve and the high-level changes needed. Key elements include:

• Purpose: The role of data in the organisation
• Scope: The scale and content of the strategy
• Future: Big-picture changes and benefits
• Objectives: Tangible goals aligned with business objectives
• Key Results: Metrics to measure success
• Capabilities: High-level capabilities to be built or enhanced

2. Case for Change

Present a compelling rationale for pursuing the data strategy. Include:

Current State: An unbiased assessment of current data and analytics capabilities

Voice of the Customer: Real-world data problems and aspirations from the customer community

Target State: Future data maturity goals tied to business needs

Business and Financial Rationale: Clear and agreed-upon assumptions on profitability and benefits

3. Detailed Data Strategy

Add detail to your strategic vision to make it implementable. Focus on:

• Future State Definition: Specific ways the future will differ from today
• Concept Designs: High-level views of future data and analytics operations
• Levers of Influence: How the strategy will be implemented and influence the organisation
• Key Gaps to Close: Documenting gaps and establishing accountability
  

4. Strategic Roadmap

Outline the phases and activities for implementing the strategy. This will include:

• 2-3 Year Roadmap: Ambitious yet flexible to account for shifts in organisational context
• Transitional States: Clear transitions to sequence the roadmap.
• Projects on a Page: High-level view of scope, activity, and resources for implementation
• Governance: Oversight, reporting structure, and governance plans

5. Target Operating Model (TOM)

Describe the people, processes, and technologies required for delivering value from data. Key components include:

• People: Roles, responsibilities, and skills needed
• Processes: Workflow and procedures for managing data
• Technology: Tools and platforms to support data initiatives
  

The 5-Step Phase: How to Create Your Data Strategy

So far, we’ve outlined what to include in your data strategy. Writing it is a big task. Breaking it down is key to success.

This section summarises how to do that and work through the process of designing a data strategy, along with important questions to consider at each stage.

Phase 1: Discover

Understand the current state and future goals of your business. Key questions include:

• What is the current maturity of your data capabilities?
• What are the experiences and needs of data customers?
• What are the organisation’s strategic goals, and how can data drive them?
• What is the vision for data and analytics?

Phase 2: Define

Outline the data capabilities that align with your organisational vision by considering the following aspects:

• Future state design principles
• Strategic questions and trade-offs shaping the data strategy

Phase 3: Plan

Establish the processes, standards, technology, and organisational changes needed. This includes:

• Strategic roadmap and transition states
• Success factors and measurement framework
• Programme mobilisation and change management

Phase 4: Execute

Delivering the transformation via iteration, gathering lessons to adjust delivery. Activities include:

• Programme management and change delivery
• Communication and training
• Project control and assurance
• Implementation monitoring and feedback loops

Phase 5: Adapt

Data strategy programmes are never a straight line – there are always twists and turns along the way. Ensuring you have the flexibiliy to meet changing demands and sustain the delivery of change. This focuses on:

• Continuous improvement aligned with key performance indicators.
• Transition and support for a sustainable future.
• Monitoring and assurance of delivery and benefits.

By following these guidelines, you can create a data strategy that effectively leverages data to successfully achieve your organisation’s strategic goals and objectives.

Don’t Just Take Our Word For It

At data experts, we know more than most that the proof is in the pudding. Visit our client case studies to learn more about the projects we’ve successfully aided.

Begin Your Data Journey Today: How Can Oakland Help? 

Wherever you are on your data strategy journey, Oakland can help you make the next step. At Oakland, we offer a range of customisable services which span the entire data strategy lifecycle.

You can choose to engage us at specific points on your journey, or we can partner with you through the entire process of defining and implementing your data strategy.

To take the first step towards leveraging your data strategy,  contact us today to chat with one of our friendly experts who can help you get started. 

Looking for more more tips and advice on all things data? Head to our dedicated knowledge hub, where you’ll find helpful guides and case studies, videos and more.  

The post How to Write a Data Strategy appeared first on Oakland.

But what exactly are the benefits of using Generative AI, and Intelligent Agents, in your business? How do you make sure you’re getting the best bang for your buck on Generative AI solutions? And how do you calculate the ROI of using Generative Artificial Intelligence effectively? 

In this blog, we’ll answer all of your burning AI ROI questions. To learn more about Generative AI , and Intelligent Agents, and their uses, explore our AI guide.

Why is ROI in Generative AI Important?

In today’s business environment you need your Gen AI to do more than provide a soundbite or provide entertainment value; you need it to deliver.  You have pressure to provide enhanced customer experiences, improve productivity, comply with regulatory demands, and drive revenue growth, and that’s before breakfast!

Which is why your senior executives want to see ROI, not R&D bills. So, if you want to unleash the power of AI, you need solutions that can do more. Generic tools or under-prepared models quickly wither in a complex business environment. At the other extreme, nobody will wait five years for a mega transformation to get the business ‘ready for Generative AI’. 

To succeed, you need the right mindset. That means building Generative AI solutions that put your business first and work with your complexity otherwise you won’t get money to do more.

Calculating the Return on Investment of your Generative AI solutions is about so much more than simply judging whether your Generative AI is worth the money you’re spending on it.

Having a clear view of your Generative AI’s ROI is one of the best ways to assess its effectiveness. It allows you to identify areas of improvement, which lets us customise your Generative AI further to suit your needs. 

A detailed view of your Generative AI’s Return on Investment is also the best way to demonstrate the value of your Artificial Intelligence initiative to stakeholders. Failing to demonstrate ROI can often hold back Generative AI projects, with early ROI proof being crucial to building buy-in and momentum for lasting change. 

Is Generative AI Return on Investment Always Monetary?

You might think of ROI as straightforward money in vs money out, but it’s far more complex than that. On top of straightforward revenue generation, the ROI of your Generative AI can be measured in:

• Enhanced operations
• Time saved
• Cost savings
• Improved decision making
• Growth
• Productivity
• Customer satisfaction
• Quality improvement.

Some of these factors, such as customer satisfaction, may be less tangible to measure. However, measuring customer retention, personalised recommendations or satisfaction survey results can all aid this. Quality improvement is also hard to quantify but can be shown through error reduction, improved accuracy or improvements in product performance. 

Ways in Which Generative Artificial Intelligence Drives Return on Investment

We’ve covered the types of ROI that AI can offer, but how does using Generative Artificial Intelligence produce these benefits for your business?

• Data analysis and insights improve your decision-making and allow you to adapt more swiftly to market changes. 
• Automating routine tasks through Generative AI frees up time for your human workers to focus on more complex tasks.
• Using Generative AI to augment your knowledge management by undertaking tasks you could never afford to resource (see our Network Rail case study).
• Generative AI can analyse customer behaviour to personalise their experiences, driving greater conversions. 
• Predictive analytics help you stay ahead of the curve and adapt to make new trends quickly work for you.
• Generative AI can use data from your workers’ CVs to create a skills matrix, allowing it to assign tasks to the most skilled person for the job. This means work is completed more efficiently.
• Using Generative AI for tasks reduces human error, preventing you from paying out of pocket for costly mistakes. 

This is just the tip of the iceberg. There are many use-cases for Generative AI. We always suggest finding a well-known problem that the business is struggling with and using this as a jumping off point. 

How to Calculate AI Return on Investment

It’s always easier to measure the ROI of your Generative AI if you know the specific goals you’re reaching for or particular areas you’re looking to improve. 

These can be quantitative, such as savings or revenue increase or qualitative, like customer satisfaction. On top of looking at internal ROI improvement in these areas, you can also benchmark them against your competitors and industry standards. 

To fully understand the progress your Generative AI is making in terms of ROI, we recommend breaking implementation down into stages to view ROI over time. This gives a detailed view, plus it allows you to see where you may have gone wrong if ROI dips at any point.

Lastly, while driving immediate ROI is significant, it’s crucial also to consider the potential for future growth and innovation.

How to Minimise Negative ROI of Generative AI

As a company, you want to protect your finances, so even if your AI consultancy has a stellar reputation, it’s not uncommon to still have reservations about spending large sums on AI platforms and solutions right away.

At Oakland, we have 40 years of making change stick. We understand large complex organisations and all of their quirks. With us, you can use smaller pilot projects or proof of concept to assess your AI’s impact on ROI before launching full-scale. That way, you can invest with confidence.

When creating your pilot projects, finding the right use cases or problems to solve is critical for driving ROI later on, as understanding value drivers of AI solutions and identifying how you will measure that early will help you set up robust value tracking from the start. Our experts at Oakland can help you with this vital step with our complementary use case workshops. 

Discover Oakland’s AI service offering today, or explore our blog to learn more about the amazing things we can do with your data. Any questions? Please contact us. 

The post How to Drive ROI with Generative AI appeared first on Oakland.

What is a Data Strategy?

Put simply, a data strategy is a long-term, detailed plan for how your company will use its data to accomplish business goals. This can include data optimisation strategies, tool implementation, the creation of new processes, and more.

Developing a corporate data strategy takes time; we know you want results now. However, managing your data without a plan doesn’t deliver the same benefits as working under a fully thought-out data strategy. 

Just give our data strategy guide a read, and you’ll soon understand why a data strategy is something you absolutely need.

What is the Purpose of a Data Strategy?

A successful Data Strategy is not about drawing a different future for your organisation or creating a standalone strategy purely on data. It’s really about explaining how data can play a key role in achieving the goals envisaged in the business strategy and what changes, investments, and capabilities are needed to make that happen.

If a business’s organisational strategy is not well documented or in flux, defining a data strategy can be a more in-depth process involving discovery and re-confirmation of organisational goals. But crucially, this doesn’t change the overall purpose of writing a data strategy.

Done properly, a data strategy will unite data-driven activity throughout the host organisation with a clear set of business goals, a compelling vision, and a case for change to drive engagement and adoption.

This quickly becomes a complex exercise covering a huge range of topics and themes, but when writing a data strategy, it’s vital to stick close to this simple purpose.

It’s data for strategy, not a strategy for data!

Why Do You Need a Data Strategy?

Until recently, many businesses didn’t consider the need for a data strategy. However, the explosion in enterprise data architecture and processing power has impelled organisations to up their game and make better data-driven business decisions.

A recurring theme is organisations skipping over data strategy and leaping straight into technical implementations. Without considering how your data management will aid your company’s ambitions, the tech and tools you choose are unlikely to be the right fit. This results in technical and business upheaval as new technology gets introduced with a localised short-term mindset instead of a longer-term strategic perspective.

If you want to succeed, it’s vital you start by thinking about the challenges and opportunities you’re facing.

Do any of these statements sound familiar?

• “There’s an opportunity in here somewhere”: Organisations often sense that they are missing opportunities to use data better but struggle to identify how. A structured data strategy design will uncover and prioritise the relevant opportunities.
• “We need to set a vision”: One powerful benefit of a high-quality corporate data strategy is that it creates a clear future to unite efforts and guide decision-making, giving organisations a competitive advantage. Sometimes, this needs to be created from scratch; in other scenarios, it’s more about helping to better structure and articulate existing thoughts.
• “Everyone’s doing their own thing”: Many organisations suffer from silos, data quality issues, disjointed data functions, and activity. Accountability may be dispersed or budgets and decision-making rights may be withheld or nonexistent. A data strategy will align priorities, resources, and roadmaps behind a clear vision.
• “Where’s the ROI?”: Either investments have been made which are not showing the expected benefits, or data sponsors and leaders are struggling to secure budget and resources to execute their ideas. The root cause is often a lack of a clear underpinning strategy.
• “The customers have had enough”: Scratch below the surface and data ‘customers’ (internal or external) are often frustrated. Availability, access rights, quality, and timeliness of data are all common frustrations. An effective data strategy harnesses these pains, identifies root causes, and mobilises solutions.
• “Our house has no foundations”: The key challenge is helping link customer frustration (“the dashboard is wrong”) to root causes (“nobody owns the data”) and solutions (“we need data governance”). This links data management best practices with a narrative grounded in day-to-day data customer challenges.
• “We’re stuck in a reactive cycle”: Many data teams become trapped in urgent demands, rework, and improvisation. This draining environment can prevent the development of long-term capabilities or enduring corporate knowledge. A collaborative process to define and implement a data strategy helps break this cycle.

Every business will have its own reasons, but if some of what you’ve just read sounds familiar, there’s a good chance you need to consider creating or updating a data strategy for your organisation.

What are the Benefits of a Data Strategy?

Still not convinced? Here are some key benefits you’ll enjoy when you take the time to implement a data strategy.

Firstly, you’ll experience enhanced decision-making. With predefined processes and easily accessible data resources, making decisions becomes more efficient and accurate. Additionally, a cohesive plan that everyone can follow minimises mistakes across the board. Optimising all processes ensures that you get the very most out of every action performed, boosting overall efficiency.

Careful planning significantly increases security and privacy, safeguarding your data assets. Future-proofing your applications means they are prepared for future advancements, keeping your business up-to-date. With a unified plan, everyone in the business works harmoniously towards the same goals, fostering team cohesion. This, in turn, improves the customer experience, increasing loyalty to your brand. Finally, by using data smartly from each sale, transaction, or deal, you can effectively optimise costs.

What’s next? Starting with the right mindset and approach is crucial.

Creating Your Company Data Strategy: The Big Picture

A key component of your data strategy should be to provide a clear view of how your different types of data will become a key engine for the wider organisation as it drives towards its business strategy. It’s where you paint the big picture of a different, data-driven future for your company.

Key elements to include in your vision are:

Purpose – What place does data have in your organisation, and why is it vital you think about it differently? Finding ways to connect this back to your organisation’s broader business purpose and strategy will be critical, as it cements the foundation of data.

Scope – What’s the scale and content of your strategy? Your scope definition should also clarify what the broader organisation understands by the term ‘data’.

Future – What are the big-picture changes and benefits you want to bring about by realising the data strategy? What will be different when the implementation has occurred? Will this improve your customer experience?

Objectives—What will we be able to achieve as an organisation as a result of the data strategy? Goals should be tangible for the business and aligned with overall business objectives. Expressing (in simple terms) how the data strategy achieves each business objective will aid the communication and buy-in of the roadmap with your senior stakeholders.

Key results – When we have achieved (or are approaching) our objectives, what metrics will we influence, and what will those look like? As with objectives, it is best to express these with key business metrics.

Capabilities – What (at a high level) will we be building or enhancing in our organisation? This section sketches out the capabilities that need to be created, developed, or overhauled.

How Does Artificial Intelligence Fit into My Data Strategy?

We have seen incredible advancements in AI in terms of how it can help business operations like decision making and generate business value with impressive return on investment. 

Bringing in AI tooling, such as Intelligent Agents or hiring a team of data scientists, won’t work as well without a solid data strategy to back it up.

Installing the latest tech buzzword tooling or hiring a team of really smart people to just “find value in the data” may seem like the right thing to do, but AI and data scientists will flounder without a good data strategy that is clear on what business problem you are looking to solve. Data is really a hierarchy of capabilities.

If you’re interested in further exploring AI’s role in your Data Strategy, see our AI guides. 

To sum up

A Data Strategy is not about giving your business a whole new set of priorities to worry about. It explains how data can play a key role in achieving the shared future in your strategy and what’s needed to make that happen.

So when you think about Data Strategy, remember: it’s not strategy for data; it’s data for strategy.

At Oakland, we’ve built an approach to data strategy development built around four key principles:

Data is business

People, Process, and Tech = Capabilities

Stories, not sermons

Co-creation and discovery

Learn more about the ins and outs of data strategy in our strategy guide, or see our blog for details of our other data services. 

The post What is the Purpose of a Company’s Data Strategy? appeared first on Oakland.

We thought we’d switch to a bit of Indie-Brit-Pop plagiarism for this blog, considering it relates to fixing things. In our previous blog, we covered the first two pain points experienced by businesses implementing their strategy for data. (In our opinion, anyway!) Today, we’ll finish the set… (more opinioning…) 

The story so far. Businesses struggle to deliver on data strategy because they fail to provide a compelling, detail-based narrative. They also don’t explain the resulting value. This translates to stakeholders not understanding the imperative for change. Which in turn leads to a lack of collaboration and co-ordination. 

TOM helps to resolve these issues by providing the detail and an achievable plan of action—the TOM package capability changes along with the arguments needed to convince and align stakeholders.  

Now, to continue the story. What are those pesky behaviours lurking that may undermine the success of your strategy? 

We promise we will learn from our mistakes… not… 

Ever worked on projects where you ask: “Can you show me how it currently works and what the issues are?” To hit a wall of blank looks and tumbleweed? In our experience, this happens more often than not. 

As the world moved to Agile, it also seems to have moved to being low documentation. Agile practitioners seem more fixated on delivering solutions than understanding existing problems. Whilst we are all for the speed and flexibility of agile delivery, the adoption of agile results in projects failing to audit the current data operating model and don’t pinpoint existing pain points that cause problems today. And it’s not that Agile can’t accommodate this activity. For some reason, it’s increasingly seen as not valuable time and effort. 

Even if you could design a ‘new’ TOM in this way, how do you show a roadmap getting you from today’s world to the future utopia? 

“Not a problem, we’ll circle back on it when we need to!” we hear this all too often. The problem is, you needed to know this before you started work on the solution. Key stakeholders need a credible design based on reality… not a punt on a wing and a prayer. Understanding “the current” is not a waste of time and effort.  

In reality, there should always be a working view of the Operating Model.  Codified and accessible as a reference resource, modified and corrected as change delivers within the organisation. If not, development of a new TOM is the perfect time to get it in place and the disciplines around maintaining it. It’ll save you a lot of time, overall. 

And the final issue?  

Well, data develops as a new data operating model silo within the business. This is where all inward-looking strategies end up in terms of their implementation. The specialist silo perpetuates regardless of best intentions on data literacy and better embedded data. And being siloed, ever growing volumes of newly keen data users are driven towards “front door” entry points, with data having to deal with ever building queues as users attempt to gain access to their vital data assets. Backlogs and workload prioritisation abound.  

Specialist data analytics remains divorced from the business. The business continues to suffer due to opaque engagement, prioritisation, miscommunication, and misinterpretation. Data becomes a fortress besieged with risk-based governance to protect everyone and everything. None of these are good outcomes. 

Target Operating Model will guide you home… 

TOM helps in both areas. Because TOM isn’t a one-dimensional document. It’s not a one-trick-pony that exclusively talks about the future. It goes into the detail of activity value chains that deliver outcomes. It talks about the capabilities and organisational components needed to underpin success. It also sets out the context of those changes in what happens today. 

To do TOM properly, your starting point should be the Current Operating Model. You gain insight on existing structures and the strengths and weaknesses of the business, its value delivery mechanisms, and resourcing model. Knowledge is power, always. And it’s enormously powerful to be able brief a stakeholder with pin-point specificity on the pain they experience. “It’s analysed. We can fix you! We know the steps, costs, and risks.” TOM provides both the blueprint and the roadmap. It shows the destination contextualised in the current. It grounds everyone in the same reality. 

Data is not just for the data geeks, everyone needs to be literate, everyone needs to take ownership of it. There’s a standardised way to define data. Data is a common feature in all areas of any business. TOM isn’t just capability, it’s culture.  

TOM helps to lay down that common approach to the management of data across the business. It should be a single approach to governance. There’s a single data classification approach. There’s a common data model everyone in the business uses. There should be an alignment to common standards. It should be the prescriptive detail that sits behind the more abstract business strategy for data. 

Data is not a silo…. It’s the circulatory system of the business. It’s the system through which critical information is communicated. It’s what provides the insight to make eyes-wide-open decisions. It provides the content to help defend the business against commercial and regulatory threats. Data needs to be connected into the organisation. Not its own organisation within the business. 

Focusing time and effort on development of a Target Operating Model that builds out the operational reality of your strategy is the way to fix you. 

 Alex Guy is a Senior Data Architect at Oakland

The post Sometimes a Target Operating Model is all the fix you need. appeared first on Oakland.

Not in our experience.  

Don’t get me wrong, the data strategy could just be ‘wrong’ but it’s a rare thing. We discussed strategy in an earlier blog here – https://www.theoaklandgroup.co.uk/a-data-strategy-is-not-a-strategy-for-data/ 

Even the best data strategies will struggle if expectations are unrealistic, or a strategy is taken as a ‘final answer’. In this scenario, strategy is seen as being something it isn’t – a magical document that will materialise into physical capabilities working collaboratively to achieve strategic objectives. It won’t and it never will. It’s the Target Operating Model that needs to stand up! 

Firstly, strategy fails because there isn’t enough traction with the group of stakeholders required to make it succeed. And generally, that’s because the change isn’t mapped out as an understandable roadmap of changes. Getting this sort of view is where the rubber starts to hit the road. 

Remember, data provides services to a customer and uses a service input from other areas. As fabulous as they are, Data Organisations are not self-contained, and they are not self-sufficient. Unless both customer and service provider understand the strategy and practical change and implications, how can they get on board with what you intend? This failure of communication and understanding is key. Customers always want better service but not if they lose what they have. Service providers would prefer to do less at lower cost and not more at the same cost.  

Understanding implication and value are essential ingredients to getting key stakeholders aligned. 

The second issue relates to “strategy for data or data for strategy”. The business’ strategy tends to cascade and generate a whole myriad of strategies. Finance, Risk, Marketing, Customer, Product, Asset, the list goes on. And each of those strategies in turn may resolve into more strategies. Leaders tend to circle around their own strategy and fend off threats to it. Data Strategy is no different and neither are its behaviours. 

It’s every person for themselves – but don’t panic it’s all going to land right and deliver. Alrighty then… 

We have a problem here! 

This is where the Target Operating Model becomes the real “north star” that’s needed by the business. Strategy is great, but winning any race is about the preparation before the race itself. This means describing the value chain of activities that deliver outcomes and the data capabilities that underpin those activities. Break your capabilities into tech, process, people skills, culture, and data features. This is where the Target Operating Model trumps the Strategy every time when it comes to executing change. Detail, specificity, prioritisation. 

Target operating model design

The early stages are all about buy-in and convincing others that your plan is credible and achievable. That means demonstrating understanding of their issues with data and how your changes will resolve them. Being clear about specific, tangible capability change helps convince stakeholders change is achievable. Being honest about the scale of change, support requirements and risk enhances your credibility. 

Laying out the principal value chains of data activity is imperative. It’s the first layer of detail connecting strategic outcomes to their physical delivery. The ‘what’.  Importantly for your stakeholders, you understand what the business is attempting to achieve and have methods that align. You also have a view on the pain they experience using ‘today’s’ methods. 

Those activities break into key organisational components: people, process, tech, and data. The ‘how’, and ‘who’ or ‘what’.  For your stakeholders, you have a handle on the more granular process details and capability maturity. Importantly, you can talk to them in terms of physical building blocks that relate to them and their issues as a customer or service provider. Remember, you have a spread of stakeholders, and they all probably have different problems to solve. 

You can provide a view on how your capability designs are arranged within the business and how that arrangement will fulfil the longer-term strategic needs of the business. The ‘when’ and ‘where’.  Linkages between different areas of capability specialism can be demonstrated. The overall impact and effect of your approach can be effectively communicated. 

And finally, you can provide a much clearer evidence-based view of the order of priority – the roadmap of change. Let’s face it, most people don’t set out on a 1000-mile road trip without some idea of the route. 

The Target Operating Model provides layered detail for discussion, challenge, and modification. Who doesn’t like a bit of shrewd discussion and bargaining? And the bargaining is worth it when it comes to “data” – good data is usually in short supply. 

Showing your method and being inclusive with key stakeholders helps solve the second issue of gaining alignment. Having a Business Strategy for data helps to make “data” an organisational effort. Everyone is involved, everyone is impacted. 

This doesn’t mean there won’t still be a series of strategies in delivery across the business, but at least the data strategy is shared. By being aligned with the areas that most influence and support it because they rely on it. Sharing is a wonderful thing, especially when everyone wins. 

Alex Guy is a design and architecture specialist at Oakland

The post Will the real Data Strategy please stand up!?  appeared first on Oakland.

Why does this happen? 

One thing failed data strategies almost always have in common is this: they were written as a ‘strategy FOR data.’ I know this sounds like an odd thing to criticise. Why wouldn’t a data strategy be a ‘strategy for data’?  

Well, the problem is this: starting with the thought ‘I’m going to write a strategy for data’ is inevitably inward-facing. Start with this, and it’s difficult to avoid slipping into: 

• talking about the technology you want to buy,  
• or explaining how the literacy of your internal user base is underwhelming 
• and/or peppering the audience with jargon 

This isn’t a great read for the folks who don’t work in data. If you start with a data-centric perspective, you will write something that’s insular, tech-heavy, and leaves the audience cold. Then, your data strategy goes on a shelf. You might be able to tick a box and say, ‘we have a strategy for data,’ but none of the expected benefits will materialise. 

It’s Data FOR strategy’ 

The point here is that your business already has a strategy, and it isn’t looking to the data team for another one. The last thing anyone needs is another department coming up with a shopping list. A Data Strategy is not about giving your business a whole new set of priorities to worry about. It’s about explaining how data can play a key role in achieving the shared future in your strategy, and what’s needed to make that happen.  

So, when you’re thinking about Data Strategy, remember: it’s not “Strategy FOR data” it’s “Data FOR strategy”.  

That is the subtle but vitally important rule to remember when you’re writing or refreshing a data strategy. 

Putting it into practice 

So far, we’ve focused on getting clear about the right mindset to bring to data strategy development (i.e. data FOR strategy). But it’s vital to translate this into an effective process for strategy development.  

Many organisations struggle with this shift. Even when you have changed your mindset, you can easily get stuck. It’s easier when you have a method you can rely on. 

At The Oakland Group we base our approach to Data Strategy development around four key principles:  

1. Data is business:  Data must be re-cast as a means to meet organisational need. At a simple level, the starting point for your data strategy is to therefore understand:
2. What outcome(s) does our organisation need to achieve?
3. What should the role of data be in that journey?
4. How will business value be created? 

People, Process and Tech = Capabilities: the modern data strategy requires a balance across technology, people and. A powerful way of looking at this is through the lens of capabilities. What does your business need to be able to do with its data? Buying technology or talent is inherently transactional and narrow in scope. Creating capability requires carefully balanced orchestration of people, processes, and technology. It opens a wider perspective for a data strategy. This is because the capabilities required will be very different depending on the needs of the organisation. If you think about capabilities as the data ‘muscles’ of the organisation, you need to know what you’re training them for. Are you running a marathon  breaking the 100m world record or strolling for the bus? 

Stories not sermons: any effective data strategy must have a compelling narrative. Writing and implementing a data strategy is fundamentally a storytelling challenge. If you can’t persuade, you can’t get anywhere. 

It’s easy to get lost in a maze of complex frameworks, jargon and detailed arguments. For the non-technical audience, a Data Strategy to tell three interrelated stories: 

The value story: how will the organisation create concrete value from data against its strategic objectives?  

The data management story: this describes how the organisation will collect, store, organise, curate and safeguard its data.  

The data culture story: this is so often overlooked, but vital. What should people think, feel, role model and advocate?  

Co-creation: too many data strategies are cooked up by a select few in the data team and then unleashed on an unsuspecting audience at short notice. These take ages to create and hardly ever land well.

A closeted approach might give the author control or make them feel clever. But it’s unlikely to get business buy-in. You’ll end up with a document that is too technocratic, too inward-facing and too peripheral to really make a difference. At Oakland, we think data strategy should be an outward-facing, whole-company effort. A data strategy is a shared document that sets out how the organisation plans to use one of its most important assets. It should grow out of workshops, customer engagement and deep analysis of the needs and personas of the host organisation. The final presentation should not be a ‘big reveal’, you MUST work collaboratively. 

When you put those four principles together, a very different way of looking at creating a data strategy emerges. One that grows as a compelling set of co-created stories grounded in the beliefs and physical reality of your organisation.  

The resulting strategy will be a careful orchestration of people, process and technology to create the capabilities needed to deliver your business’ strategy. It will be a lot more than a ‘strategy for data’. 

Joe Horgan is a Principal Consultant at Oakland, he regularly posts on LinkedIn about all things data strategy and digital transformation. Follow him here….https://www.linkedin.com/in/joe-horgan/

The post A data strategy is not a ‘strategy for data’  appeared first on Oakland.