AI in Your Business: Start With a Practical Pilot
Your team has heard plenty about artificial intelligence. The harder question is where it could make a useful difference in your company, and how to introduce it without creating avoidable risks.
A sound AI project starts with the work itself. I help businesses move from an initial conversation into a focused audit. Then we run a practical pilot and evaluate its results so decision-makers have evidence for what to do next. GDPR and EU AI Act considerations shape the process from the beginning. Staff training is part of the plan as well.
Begin with a business problem
It is easy to start with a product demonstration and then search for a task to fit it. A better starting point is repetitive work that slows a team down. For example, employees might spend hours sorting incoming requests or drafting routine responses.
Those examples show what we can investigate. Results depend on the task and its data. A process may look suitable for AI and still depend on information that is incomplete or sensitive. Sometimes access is the obstacle. The first conversation clarifies the goal and the constraints that shape a useful next step. We also identify who may be affected.
The initial consultation is free. We can discuss your questions, the processes you have in mind and whether a more detailed review would be worthwhile.
An audit shows where AI may fit
If there is a promising direction, the next step is a detailed audit. Together, we map relevant workflows and systems. We also look at who would use the solution and who could be affected by its output. Then we define what a reliable result would mean in practice.
The audit distinguishes AI opportunities from tasks that may suit process changes or conventional automation more closely. Where AI seems suitable, we define a contained use case and the conditions for testing it.
The audit can also produce a shortlist for management. Each candidate can be described by the problem it addresses and the information it would need. The audit can also explain how the idea would change people’s work. That makes it easier to compare a promising experiment with the effort required to support it. Some ideas may depend on cleaner records or a revised workflow before a pilot can begin. Identifying those prerequisites early helps set a realistic order for the work.
That review also creates a place to raise governance questions early. Depending on the proposed use, the company may need to examine personal data, access permissions, retention and the risks of inaccurate output. The GDPR sets principles for processing personal data, including data minimisation and security. Their practical application depends on the data and the specific processing involved. The GDPR text on EUR-Lex is the primary reference.
Test one use case before expanding
A pilot turns the audit’s findings into a small, observable test. Before it begins, we record how the task currently works. That baseline gives the team something concrete to compare against when the trial ends. We can then assess the time saved and whether the output meets an agreed standard. We also check how well the system fits the existing process. The measures depend on the use case. A drafting assistant, for example, may be judged by the quality of its first drafts and the amount of editing employees still need to do.
We agree what the system should do and what information it can use. We decide where a person must check its work, then set measures for evaluating the results. The pilot could support a staff member preparing an answer, while the employee remains responsible for reviewing and sending it.
Evaluation matters as much as setup. We look at whether the pilot helps with the chosen task, where its answers fall short and what extra work it creates for staff. We also pay attention to edge cases. A system that performs well on routine requests may need a clear handoff when a request is incomplete or unusual. A useful outcome may be a case for wider implementation. It may also be a clear reason to adjust the process or stop the test.
Once the pilot has been evaluated, I can assess further AI opportunities across the company on request. That gives management a more grounded basis for prioritising the next project.
Build GDPR and AI Act questions into the work
Compliance questions are easier to address while a use case is still being designed. A practical review can identify what data enters a system and which provider handles it. It also checks what contractual and security information is available, then considers how people can review important outputs. Some situations call for input from a data protection officer or legal counsel, especially where the processing or potential impact is significant. Vendor terms and data handling practices deserve close attention. A tool’s public-facing features may say little about where submitted information is stored or how it may be used. Check those details before business data enters the system.
The EU AI Act uses a risk-based framework, and its provisions apply in stages. The European Commission says AI literacy provisions have applied since 2 February 2025; other obligations have different dates and depend on the system and its role in the supply chain. A company should therefore assess each intended use instead of assuming every AI tool follows the same obligations. See the Commission’s AI literacy questions and answers and its AI Act overview.
This work helps a business make informed choices and prepare appropriate safeguards. The company retains responsibility for compliance, and some questions call for legal advice.
Give leaders and employees the training they need
Management needs a working understanding of AI’s capabilities and limits. Training can help leaders ask better questions about proposed projects. They can assess pilot evidence and understand likely costs. Training also helps them see where human oversight matters.
Employees need guidance connected to their day-to-day work. They should know which tools are approved, what information they may enter, how to review generated content and what to do when an answer seems wrong. Practical exercises based on familiar tasks make those expectations easier to apply.
The European Commission’s explanation of AI literacy says organisations should take measures to support the AI literacy of people who operate or use AI on their behalf. It says to take account of staff knowledge and experience, along with the context of use. Training belongs in the introduction plan, with guidance ready before people begin using the systems.
Take the first step with a conversation
A complete AI strategy can come later. Bring a recurring task or a question from your team to a free initial consultation, and we can decide whether it deserves a closer look. From there, we can decide whether an audit and pilot make sense for your business.
