How to Build an AI-First Engineering Team in 90 Days

The Challenge

When I joined Mindvalley as a senior engineer, the team was talented and productive — but AI felt like something that happened at other companies. We shipped great software, but we weren't leveraging the tools that were rapidly changing what was possible.

Over 90 days, we shifted that. Here's the exact playbook.

Why Most "AI Transformation" Efforts Fail

Most companies approach AI adoption one of two ways — both of which fail:

1. Top-down mandate with no support. Leadership announces "we're an AI company now." Engineers are confused about what that means and afraid of being replaced. Nothing changes.

2. Bottom-up experimentation with no direction. A few enthusiastic engineers build demos and POCs. They're never productionised. The rest of the team doesn't engage.

The right approach is neither: it's a structured, opt-in programme with clear milestones and business outcomes.

Week 1–2: Create Psychological Safety

Before anything else, address the elephant in the room: fear.

Engineers worry that AI will replace their jobs. If you don't address this directly, you'll have silent resistance.

In our case, I ran a 90-minute session called "AI as leverage" that reframed the conversation:

• AI is a multiplier on engineering output, not a replacement for engineering judgement
• Teams that adopt AI tools will be more productive and more valuable — not redundant
• Our goal is to ship better products faster, not to reduce headcount

This sounds obvious, but saying it explicitly — and having leadership echo it — made a measurable difference in team receptiveness.

Week 3–4: Identify Your AI Champions

You don't need the whole team to adopt AI at once. You need 2–3 engineers who are genuinely curious and willing to experiment.

These aren't necessarily your most senior engineers. Look for engineers who:

• Ask "why" questions a lot
• Have side projects involving new technology
• Are comfortable with ambiguity

Give your AI champions protected time (20% of their sprint) to explore AI tooling and bring back findings to the team.

Week 5–8: Find One Real Problem to Solve

The fastest way to change culture is results. Pick one real, frustrating problem the team faces and solve it with AI.

At Mindvalley, we started with code review comments. We were spending 2–3 hours per week on repetitive, style-related review feedback that could be automated. We built a lightweight AI-powered PR reviewer that caught common issues automatically.

The impact was immediate and visible: engineers got faster feedback, reviewers spent less time on boilerplate, and the AI champion who built it got visible recognition.

The key: pick a problem where success is measurable and visible to the whole team.

Week 9–12: Standardise and Scale

Once you have one success story, create a lightweight playbook:

1. A short list of approved AI tools (avoid tooling sprawl — pick one LLM API, one code assistant, one vector database)
2. A template for AI integration (how to add an LLM call, how to write a prompt, how to evaluate output quality)
3. A monthly "AI demo day" where engineers share what they've built

The monthly demo day is the highest-leverage activity in this list. It creates social proof, surfaces good ideas across the team, and rewards engineers who experiment.

What Not to Do

• Don't mandate specific tools. Evangelise and support, don't force.
• Don't start with the hardest problem. Start with a quick win. Complex problems require mature AI practices.
• Don't skip evaluation. Every AI feature needs a way to measure whether it's working. "It seems to work" is not a production standard.
• Don't ignore security and data privacy. Before any customer data touches an LLM API, involve your legal and security teams.

The Results

After 90 days at Mindvalley, we had:

• 4 AI-powered features in production (code review, content tagging, support ticket triage, internal search)
• 80% of engineers using AI coding assistants daily
• A monthly AI demo day with consistent attendance from senior leadership
• A playbook that onboarded new engineers to our AI practices in their first two weeks

None of this required hiring AI specialists. It required structure, psychological safety, and one early win.

---

Building an AI-first team and want an outside perspective? Book a strategy call — I've done this at multiple companies and can help you avoid the common pitfalls.