10 questions we'll ask before undertaking an AI project

Implementing AI is never just a tech decision. It’s a business decision with legal, ethical, and operational implications. That’s why before we write a single line of code, we work with our clients to answer these 10 questions.

1. What’s the real business goal?

We start with “why”. What outcome are you aiming for? Are you trying to automate a task, improve decision-making, personalise experiences, or something else? AI is a tool, not a strategy. It only makes sense when it’s solving a business problem that matters.

2. What does success look like?

We ask clients to define clear and measurable expectations: How accurate does the system need to be? What’s the acceptable margin of error? What’s the threshold where it stops being helpful or safe? Getting this right early avoids misaligned delivery later.

3. What kind of data do you have, and is it enough?

We assess the quality, volume, and representativeness of your data. Are there gaps? Are both positive and negative scenarios covered? Can we validate outputs against a trusted source? If not, do we need a manual review process in place?

4. Can your data be used, shared, or moved?

Some organisations can’t allow data to leave their environment. We ask: Are there hosting constraints? Do you need containerised services? These choices influence the architecture and feasibility of a project.

5. Are there regulatory, compliance, or privacy constraints?

We surface obligations up front: Is this project impacted by health data laws, financial services regulation, GDPR-style privacy requirements, or internal governance? These factors will shape what can be built and how it must be deployed.

6. Have ethical risks been identified, and who’s responsible?

We ask who is accountable for the ethical use of the model. That includes bias, explainability, transparency, and unintended harm. Ethical AI isn’t just about fairness, it’s about reducing organisational risk, legal exposure, and reputational damage.

7. What’s your current experience with AI?

Knowing a client’s maturity helps us tailor the approach. Are you just starting out or do you already have a data science team? This changes how we engage, what kind of support is needed, and how we frame delivery milestones.

8. Who is monitoring the model after it goes live?

AI projects don’t end at deployment. Models degrade. Business contexts change. We ask how outputs will be monitored over time and who will be responsible for acting on issues before the model causes problems for customers or the business.

9. Who’s going to support the AI solution over time?

We ask who’s going to own the model once it’s live. AI systems need ongoing maintenance. That includes updates, retraining, performance checks, and user feedback loops. Without a clear owner, models quietly degrade or get sidelined.

10. Is AI the right solution?

Not everything needs AI. Sometimes, a simple rules engine, workflow automation, or analytics dashboard delivers faster results with less overhead. Just because AI can be used doesn’t mean it should.

Why this matters

By asking these questions early, we:

• Reduce risk
• Set realistic expectations
• Align teams
• Avoid costly rework
• Build better AI solutions.

Ready to explore AI with clarity and confidence?

Let’s talk about how these questions apply to your organisation. Whether you're just starting out or scaling existing AI initiatives, we’re here to help you build smarter, safer solutions that actually deliver value. Start the conversation today.