It’s well known that GenAI has rapidly emerged as a transformative technology and harnessing its potential is critical to stay competitive. But how do you start your GenAI journey?
Mark Rotheram, Chief Technology Officer at BCN, shared his insights during our Digital Summit and outlined a practical roadmap for businesses eager to get started with GenAI and Microsoft’s Copilot.
Here are Mark’s five essential steps to start your AI journey successfully.
1. Set clear objectives from the start
Before jumping into the GenAI space, you must fix on a clear objective. One of the most common pitfalls businesses face is starting without a defined purpose. AI technologies are powerful, but they’re only as effective as the problems they’re designed to solve. Do you want to automate routine tasks? Improve customer engagement? Or perhaps enhance productivity in document management?
It’s essential to identify your key business challenges and focus on how AI can address them. An example might be using AI to automatically process invoices from emails – a relatively simple task, but one that could save considerable time and effort. By starting with clear, measurable goals, you can align your AI initiatives with your business needs and avoid wasting resources on low-impact applications.
2. Evaluate your current capabilities
Next on the list is to take stock of your organisation’s existing capabilities, including your technology and your people. This is crucial in determining whether your AI initiative will thrive. Do you have access to AI and machine learning specialists? Do your teams have the necessary data literacy to interact effectively with AI tools? If not, training or hiring might be necessary prior to starting.
My advice is to start small and lean into the tools you already have. Many businesses, for example, already use Microsoft 365, which includes access to Copilot, the AI tool that integrates across Microsoft products like Word, Excel, and Teams. Leveraging tools your employees are familiar with ensures a smoother transition into AI-driven workflows.
Additionally, explore whether it’s better to build or buy your AI solution. Depending on the complexity of the task, you may find that off-the-shelf solutions, such as Copilot, meet your immediate needs. For more specific challenges, however, you might consider building custom models using Azure OpenAI, which offers a higher level of customisation but requires a more significant investment.
3. Prioritise data quality and security
One of the most important elements of any AI initiative is data. Good data is the foundation on which useful AI models are built, yet many organisations overlook the importance of structuring their data properly. Businesses need to ensure they have access to high-quality, relevant data that can be fed into their AI models.
However, the use of data comes with a privacy and security warning. Many AI solutions, particularly public ones like OpenAI’s GPT, come with risks regarding where your data is stored and how it is used. For businesses in sensitive industries or those handling proprietary data, it’s important to choose AI platforms that ensure your data stays secure. Solutions like Microsoft’s Copilot provide built-in safeguards, making them an ideal choice for organisations concerned about data privacy.
4. Select the right AI tools for your use case
With clear objectives, capabilities assessed, and data in place, the next step is to select the appropriate AI tools. There are broadly four tiers of AI implementation, starting from simple reactive tools to more complex, custom-built solutions:
Transformed: The final stage involves custom-built AI solutions that fundamentally transform the way your business operates. These might include bespoke applications that process unstructured data or sophisticated chatbots trained on your company’s data. Such initiatives require deeper investment but can offer significant competitive advantages.
Reactive: Many businesses already have access to basic AI tools embedded within their software subscriptions. For example, Microsoft 365 includes a free version of Copilot, which can assist with simple tasks like drafting emails or organising documents. This is an excellent starting point for organisations just beginning their GenAI journey.
Foundational: A step up from the free version, the paid M365 Copilot unlocks more advanced functionalities like integrating with Teams for meeting transcription and note-taking, or generating automated reports from Excel and Word. These tools provide a more comprehensive way to streamline everyday business processes.
Enhanced: This level involves embedding AI directly into your core business systems, offering tailored solutions for more sophisticated tasks, such as integrating AI into your sales pipeline or customer service operations.
5. Monitor, refine, and iterate
The AI journey doesn’t end once you’ve deployed a solution. In fact, the most successful AI projects are those that are continuously monitored and refined. My advice is to set up clear processes for measuring the effectiveness of AI initiatives. Are you achieving your initial objectives? What insights can you gather from the data? Are there areas where the AI is underperforming? Regular reviews ensure that your AI applications are delivering value and can adapt as your business evolves.
A key part of this step is adjusting your AI’s training as it learns from new data. For example, BCN’s AI solution for invoice processing improved over time by learning from the company’s feedback and refining its accuracy in data extraction. The more you iterate and improve, the greater the long-term value you will derive from your AI investment.
