We’ve all seen the AI hiring scramble. And while there’s value in bringing in specialists, there’s also something quietly powerful about developing what you already have.
Recently, I had the chance to support a team of developers who transitioned into building real AI features, not in years, but in under 6 months. It wasn’t flashy. It was structured, thoughtful, and surprisingly effective.
What worked:
- 6 weeks of focused AI/ML training tailored to real-world needs
- 3 months of internal mentorship through lightweight “AI labs”
- A series of small, scoped projects that turned theory into traction
- Occasional expert input, brief, but impactful (architecture, review)
This aligns with what I’ve seen in teams that scale AI well:
- Start with a clear skills inventory
- Personalize the learning path, not everyone learns the same way
- Keep the ratio practical (60% doing, 40% learning)
- Treat delivery projects as opportunities to learn, not barriers to it
The outcomes surprised even the team:
- Faster than hiring (2x shorter ramp-up)
- Higher retention and morale
- Stronger integration of AI with their existing domain context
People sometimes ask: Should we build AI capacity internally or hire for it?
It depends. But I’ve found that when you have thoughtful, steady engineers who already care about the product, they can often grow into this role, especially with the right scaffolding.
A principle I try to design around:
Capability sticks when it’s built in context.
Hiring brings skill. Internal growth brings alignment.
