The first step to adopting AI is the simplest and most frequently overcomplicated part of the entire process. You do not need a strategy document, a technology audit, or a budget approval. You need to identify one specific task that takes real time and test an AI tool on it. The BCC found that 71% of UK SMEs have not identified a need for AI, which means most businesses are stuck before they even start.
This guide explains why the first step matters, how to choose the right task, and what to do after your first test. Everything in AI adoption flows from taking this first practical action.
Why Is Task Identification the Real First Step?
Many businesses start AI adoption by evaluating tools, reading about models, or writing strategy documents. These activities feel productive but do not generate evidence. The first step that actually matters is identifying a specific task where AI could save time or improve quality.
This matters because AI is a tool, not a strategy. Without a specific problem, you cannot evaluate whether any tool is working. The BCC survey showing 71% of SMEs lacking an identified need is not about technology, it is about not knowing where to point the technology. Once you name a task, everything else becomes concrete: which tool to try, who should lead, and how to measure success.
How Do I Choose the Right First Task?
The best first task has four characteristics. It is repetitive, meaning it happens regularly. It is time-consuming, taking at least a few hours per week. It involves processing text, data, or documents, which is what current AI tools handle best. And it has a human in the loop, meaning someone reviews the output before it matters.
Tasks that fit this profile include drafting customer email responses, summarising long documents or meeting notes, generating first drafts of marketing content, extracting data from invoices or receipts, and creating meeting agendas from transcripts. Tasks that do not fit include anything requiring perfect accuracy without human review, decisions with significant financial or legal consequences, and tasks involving sensitive personal data without proper safeguards.
For a complete framework, see our guide on identifying AI use cases.
Should I Write an AI Strategy First?
No. This is one of the most common mistakes we see. Businesses spend weeks or months writing AI strategies before anyone in the company has used an AI tool for a real task. The resulting strategies are generic because they are not informed by practical experience.
The better sequence is: try AI on one task, learn what works and what does not, then write a strategy that reflects your actual experience. A one-page strategy written after a successful pilot is worth more than a fifty-page strategy written before one. McKinsey's research on AI adoption confirms that organisations with hands-on AI experience make better strategic decisions than those that plan extensively before doing.
Who Should Lead the First AI Project?
Assign the first project to the person closest to the task, not to IT or senior management. The ideal candidate is someone who currently does the task, is curious about AI, and is motivated to find efficiencies. They do not need technical skills, they need willingness to experiment.
Give this person time to experiment, access to a tool like ChatGPT Plus or Microsoft Copilot, and a clear expectation: try AI on this task for two weeks and report back. Avoid assigning the project to someone who is sceptical or overburdened, as their experience will colour the outcome. For guidance on building team capability, see what skills your team needs for AI.
What Happens After the First Step?
After your first test, you have one of two outcomes. If the AI tool saved time or improved quality, you have evidence to support expanding. The next steps are to document what worked, share the approach with other team members, and identify a second task to try. If the tool did not help, you have learned that either the task or the tool was wrong, which is equally valuable.
The key principle is momentum. Each test, whether successful or not, teaches you something about where AI fits in your business. Businesses that run sequential small pilots learn faster than those that wait for the perfect plan. The BCC data showing only 20% of small businesses seeing AI as accessible suggests that perceived barriers are higher than actual ones once businesses start experimenting.
If you want help taking that first step, book a free discovery call. We help UK SMEs identify their first viable AI task and set up a pilot that produces real evidence. See our services for the full range of support we offer.