The pace of change in our industry is relentless. New AI-enabled tools show up every month, existing tools shift under our feet, and it’s nearly impossible for any one person to keep up. At the same time, our customers are facing a strategic choice: tap into the clear upside of AI—smaller teams moving faster with far less manual effort—or wait until questions about code quality, IP, compliance, and compatibility with legacy environments feel more settled.
While frontier companies have already made the leap and are building with AI every day, many others are still somewhere between curious and cautious. That leaves our product and engineering teams in a bind: we need to stay fluent in tools our customers may not yet be ready to adopt—and we need to be able to separate hype from reality so we can guide them with clarity. That kind of fluency comes from regular use, shared learning, and a willingness to experiment together, not from individual side projects.
To make this a habit, we carve out time on the calendar for everyone to work with AI tools in a safe, sandboxed environment. No client code, no pressure—just focused exploration. Yesterday’s focus was revisiting prototyping tools to see what’s changed in the last few months.
How do you structure focused, low-stakes AI learning for your team?
We gave the team a simple constraint: spend an hour prototyping with either V0 or Figma Make. The work was mandatory; what people built was up to them. Some opted for the provided prompt to remove barriers to getting started and minimize context-switching costs, while others used the time to test ideas they’d been simmering on: improving a planning poker app, exploring multiplayer experiences, experimenting with consumer integrations, etc.
People moved fast, tried ideas, and got feedback in real time. Everyone ended the hour with an updated sense of what’s possible—and clarity on how to bring these tools into client work when the client is ready.
What were our latest learnings?
Figma Make still works best with a single example frame.
Connecting a full component library is improving, but prompting with a clean frame still produces the most consistent, on-brand output. This matters, especially since most companies already have Figma in their procurement-approved stack.
V0 is the leader when rapid prototyping with a backend.
The team explored the improved native Supabase integration and also tested connecting via API key with Todoist (Python backend services). Both were quick and seamless to set up, whereas Figma Make generated a stateless prototype with temporary storage.
Shared learning beats solo experimentation.
Time-boxing the effort and bringing everyone together created space for cross-pollination—shortcuts, patterns, pitfalls, inspiration. The conversation after the hour was as valuable as the builds themselves.
Why this matters for our customers
Whether it happens this year or next, every client will eventually shift how they build software. AI-assisted development won’t be a novelty; it will be the expectation. And when they do make that shift, they’ll need partners who:
- understand the evolving tool landscape,
- can move quickly without cutting corners,
- know how to use AI responsibly,
- can evolve early prototypes into enterprise-grade, production-ready solutions, and
- meet them with empathy as they navigate the transition.
Protecting dedicated time for learning -- alongside day to day work with teams who have already embraced AI tools as part of their working model -- helps ensure our team is already there—skilled, confident, and able to guide clients with real-world experience instead of theory.