Integrating AI into Your Teams: What You Really Learn by Doing It
How a Practical, Tooled-Up, and Progressive Innovation Culture Can Make Your Company Much More Competitive
When we talk about AI in companies, there's often a disconnect. On one side, the excitement—or pressure—of leadership. On the other, the skepticism or hesitation of some team members. And that’s perfectly normal.
At Tech2Heal, and now in my consulting work, I see this tension every day. AI intrigues, but it also unsettles. However, those who truly integrate it into their way of working (beyond an isolated POC) gain speed, quality, and engagement—provided they do it in a structured and pragmatic way.
Why It Matters Now
AI doesn't replace teams. It redefines roles.
For us, a product spec no longer takes three days of back-and-forth to write and refine. It starts in 30 minutes with GPT-4o, then gets enriched and reviewed by a PM, and is directly used in tools like Stitch or Locofy to generate a visual prototype. The design takes shape almost instantly. Yes, we generate usable HTML or Figma outputs.
It’s not magic. It’s a new set of habits. And it changes everything.
If you're wondering whether your company is ready for this shift, read Is your company ready for AI? to assess your current maturity.
1. Train Without Pressure—But Train Anyway
Many teams wait for “the right AI,” “the perfect tool,” or for IT to approve everything before trying anything. Result: nothing moves. The trick? Train gradually.
At Tech2Heal, we started with short sessions (30 minutes) to show how developers could use Cursor or GitHub Copilot to speed up tasks. Not to replace them, but to help avoid starting everything from scratch.
Same for design: using Stitch or Galileo to turn a text spec into a clickable wireframe is a shock at first. But once teams see the time savings, skepticism turns into curiosity.
2. Set Up a Shared and Ongoing Watch
Innovation shouldn’t stay in the hands of leadership or an “innovation committee.” We set up a collaborative watch loop with the following roles:
- One AI referent per team (product, dev, data, design)
- A monthly review of the best tools tested
- A shared knowledge base (e.g., Notion) where anyone can document trials
This creates a natural momentum. Tools evolve fast. But if you structure your monitoring, your company keeps up—without depending on a guru or external consultant.
3. Let Non-Developers Build Prototypes
This is a game-changer: today, with tools like Lovable, a non-tech person can build a working UI component that developers can later integrate.
I’m not saying the generated code is perfect. But it saves time. And more importantly, it builds bridges between teams. A PM who can deliver an initial HTML mockup is more engaged, more credible, and improves communication with devs.
On a client project, we halved the front-end production cycle for a new feature by applying this principle.
4. Accept That AI Isn't “Perfect” — But It's Still Worth It
A common objection: “AI sometimes gets it wrong.” True. But it’s not a bug—it’s the deal.
Every time you use an AI assistant, you save time if you know how to review, correct, and refine. It requires upskilling—not just technically, but in critical thinking.
Our approach: we treat AI like a brilliant but inconsistent junior. It drafts the first pass. You revise. Over time, teams learn how to write effective prompts, extract reusable code blocks, or generate automated test feedback.
5. The Developer Role Is Evolving—Prepare Them for It
In many teams, developers are still seen as “the ones who write code.” But this role is evolving: today, they’re increasingly assemblers, supervisors, and connectors. Their value no longer lies solely in writing everything from scratch, but in understanding, orchestrating, and integrating tools—often powered by or accelerated by AI.
In practice, a developer who knows how to use an AI assistant like Cursor or Cline, who can adapt generated code, or oversee a UI built in Lovable, can move 2–3x faster on some tasks. And this isn’t theory—we see it in our processes, both at Tech2Heal and in consulting missions.
Of course, this shift requires cultural change. Some developers—often the most experienced—may resist: “this code is messy,” “it’s unmaintainable,” “it’ll never go to prod.” And they’re sometimes right. But the point is: this isn’t an end state, it’s a starting point.
The real challenge is giving them the framework, tools, and permission to work differently. To say: my job isn’t to do everything myself, but to ensure it works, it holds up, and it moves fast. It’s not about “replacing” developers—on the contrary, AI makes them more essential than ever, as long as they let go of the need to control every line.
In our case, we’ve started running internal sessions on these topics: how to use an AI assistant effectively? How to review generated code? Where do we draw the line between automation and technical debt? These discussions shift mindsets. And they save time—without sacrificing quality.
What I Would Do Differently
At first, we thought the main barrier would be technical. In reality, it was cultural.
We should have worked on change acceptance earlier. Better explained the gains. Highlighted small, concrete wins. And above all, celebrated the internal pioneers.
A developer testing Lovable or Cursor on their own time deserves a space to share. They are the real drivers of change—not the slides.
In Conclusion
Integrating AI into a company isn’t something you decree. It’s something you build. And it often starts with simple steps: train, test, share.
But if you do it right, you become faster, more creative, more attractive. And you prepare your teams for the future of work.
If you’re working on structuring a real AI strategy for your company—or wondering where to start—feel free to reach out. I love challenging real-world use cases and would be happy to discuss.
Further Reading
If this approach resonates with you, also check out:
- Why Embracing a Tech & Innovation Culture Matters – The strategic importance of this transformation
- Is your company ready for AI? – The essential prerequisites before diving in
- Avoiding Costly AI Prototypes – How to validate your AI ideas quickly and cost-effectively