Key takeaways:
- Engineering teams adopt AI faster through peer-driven learning than top-down mandates. Shared pilot stories create pull, not pressure.
- New development paradigms, in which teams write clear intent and acceptance criteria before using AI to generate options, are emerging as a high-value workflow for AI-assisted software development.
- AI isn’t replacing software engineers, but expanding who can participate in building, accelerating feedback loops, and creating new opportunities for cross-functional collaboration.
As SVP of Development, I’ve lived through plenty of platform shifts, but AI feels different. It’s a compounding advantage that shows up everywhere: how we clarify requirements, how we explore designs, how we prototype, how we write and review code, and how quickly we learn our way through ambiguity.
Early on, we kept coming back to a simple question:
If there were five of you, and time was no object, what would you build? What would you fix? What would you finally make possible?
This has been the year of dreams. The change came fast, but not all at once. Looking back, I see three themes that turned AI from curiosity into a real transformation for our teams: inspiration, pilots, and evolving roles.
1. Inspiration: How We Built an AI-First Engineering Culture, Without Mandates
Our shift didn’t start with an executive directive. It started with engineers being curious and experimenting. One very experienced team member told me he was skeptical at first. He assumed AI would only be useful for generating brand-new code. So he pointed it at an item on our roadmap, fully expecting AI to fail. Instead, he came away impressed with how much it delivered.
Our teams love to win and love to learn, so those early stories created the right kind of pull: “you’ll fall behind if you don’t at least try it.”
That dynamic — curiosity leading to a result, a result leading to a story, a story spreading through the team — turned out to be more powerful than any top-down AI initiative could have been.
2. Pilots: What 20 AI Pilots Taught Us About Software Development at Scale
Next, we encouraged teams to pilot spec-driven development: start by writing intent and acceptance criteria clearly, then use AI to generate options (design approaches, scaffolding, tests, and first-pass implementations) before committing to a direction. The idea is to front-load the thinking, so AI is amplifying a clear human intent.
We expected to run 3–5 pilots. We ended up running closer to 20.
It was a bit chaotic – and that was okay. We set up a weekly sync to share learnings and to give someone the spotlight to tell a specific story: a problem they solved faster, how they made AI work more effectively, or a lesson they learned the hard way so others didn’t have to.
No one person has all the answers, but together we learn quickly and scale what works.
If you’re thinking about how to structure AI adoption in your own software team, our biggest unlock was giving people permission to fail publicly — and making it easy to share what they found on the other side.
3. What’s Evolving: How AI in Software Development Is Changing Roles and Who Gets to Build
One of the most encouraging shifts has been how much AI has democratized the work of software development, and how quickly roles are blending. More people outside of engineering are getting comfortable prototyping and validating ideas earlier. That translates into better communication, faster feedback loops, and better decisions.
At Precisely, this has looked like product managers generating rough prototypes to pressure-test a concept before it reaches a developer. It’s looked like data teams scaffolding internal tooling they’d previously have had to wait months to prioritize. The bar for “I can build something to test this idea” has dropped significantly, and that’s a good thing.
People have taken that “dream big” prompt and run with it. From tackling major re-architecture projects, to creating new product concepts in record time, to building internal tools that free up hours each week. In upcoming posts, we’ll feature a few of these teams and what they learned along the way.
What This Means for the Future of Software Engineering
AI isn’t replacing the craft of software development but instead changing the leverage we have when we apply that craft. My goal is to make sure we use that leverage to build better products, create more opportunities for our teams, and stay focused on outcomes that matter.
The engineers and builders who thrive in this environment bring clear thinking, creativity, strong judgment, and a willingness to share what they discover. That’s the kind of team we’re building at Precisely.
If you’re on a similar journey, I’d love to compare notes.
Frequently Asked Questions About AI and Software Development
How is AI changing the way software development teams work?
AI in software development is shifting teams from linear, sequential workflows toward more exploratory, iterative ones. Rather than writing all requirements upfront, teams can now use AI to rapidly generate design options, test implementations, and scaffolding — then evaluate and refine. The biggest cultural change is that learning happens faster and spreads more easily when teams share pilot results openly.
What is spec-driven development with AI?
Spec-driven development is a workflow in which engineers write clear intent and acceptance criteria before engaging AI tools. By defining the goal first, teams get more useful AI-generated options — whether that’s code scaffolding, test cases, or alternative design approaches — and make better decisions about which direction to pursue.
How do software leaders drive AI adoption without top-down mandates?
The most effective AI adoption tends to start with voluntary pilots and peer storytelling. When one engineer shares a result that surprised them, others want to try it. Leaders can accelerate this by creating structured forums — a weekly sync, a shared channel, a recurring spotlight — where teams share what they learned, including what didn’t work.
