Generative AI isn’t just accelerating code. It’s changing how startups build — and who builds. It’s rewriting the rules of what it means to be an engineer, a product manager, and a high-performing team.
Today, working prototypes are being produced without involving engineering. Junior developers are using GenAI tools to produce senior-level code. And top CTOs are rethinking who they hire, how they hire them, and what excellence even looks like on a modern team.
For founders and builders, these shifts aren’t theoretical — they’re existential. Whether you’re building with five engineers or fifty, GenAI is already forcing decisions about who to keep, who to coach, and who to hire. Every builder and founder is facing a high-stakes shift: either re-architect your team for a GenAI-native world or risk being outpaced by competitors who do. This isn’t about tools. It’s about a fundamental reset of roles, expectations, and the very definition of velocity.
The Old Pyramid Is Starting to Tilt
The first and most immediate shift is structural. The classic engineering org — built like a pyramid with layers of ICs and middle managers — is shifting under the weight of GenAI. Leaders aren’t just adding AI to the stack; they’re rethinking the composition of the team itself.
During Madrona’s Builders Summit in May, Kong Head of Engineering Saju Pillai described his team, whose product handles approximately twice the monthly network traffic of Netflix. It’s a systems-heavy org, where mistakes are expensive and performance and reliability are gospel. Yet even in that high-consequence environment, he sees AI reshaping who belongs. His prediction? The traditional broad base of junior and mid-level engineers at the base of the pyramid will steadily narrow, leaving a leaner structure focused on high-leverage talent.
At Talkdesk, CTO Munil Shah framed the shift more bluntly: A modern team needs only two kinds of developers: god developers and supervisor developers. His vision of this “supervisor” developer is one who is fully proficient in using AI tools and we all know what he means when he refers to “god developers.” The job of all the other developers is to prove they belong in one of those two buckets.
For founders building technical teams, this clarity is useful. This isn’t a call for mass reductions. Instead, it’s a call for sharper focus: build teams around the kinds of engineers who amplify GenAI’s strengths and multiply velocity.
The tactical takeaway: Audit your teams for who is advancing GenAI-driven velocity and who is passively watching it. Use those signals to reshape hiring profiles and internal growth paths — but do so with intention.
Don’t Just Hire Coders — Hire Contextual Thinkers
Once you’ve redefined what kinds of engineers your team needs, the next challenge is figuring out how to identify and evaluate them. GenAI doesn’t just change who you hire — it transforms how you hire. Traditional coding interviews focused on syntax and algorithms are increasingly out of step with the reality of AI-assisted development. What matters now is not whether someone can write code in isolation, but whether they can solve complex problems, think critically, and use GenAI tools as creative collaborators. The best engineers aren’t just using these tools — they’re shaping how those tools operate. They’re teaching the AI what matters, embedding context, judgment, and nuance into the prompts and systems they build.
It is time to tear up the interview rubric altogether. Talkdesk, for example, now splits interviews 50/50 between traditional technical questions and open-ended problems where candidates are expected to use any tool — GenAI included — to reach a solution. If a candidate is not using these tools, they simply won’t make the cut.
Why? Because the ability to use AI tools isn’t the ceiling — it’s the floor.
The real differentiator is critical thinking. Can someone decompose a business problem into solvable chunks? Can they design a system that aligns with customer context? As Saju from Kong emphasized, “Even the act of decomposing the problem into the right pieces… that comes with experience. And it’s an art.”
The tactical takeaway: Rethink your hiring loop to balance problem-solving and strategic thinking with coding. Create space for product sensibility and architectural thinking. Ask candidates to solve open-ended problems using the same GenAI tools they’d use on the job. Evaluate how they reason, what trade-offs they consider, and whether they can harness AI not just to code, but to think. Reward those who can teach the AI what matters, not just use it.
Product Is Everyone’s Job Now
In this new world, code writing is no longer the bottleneck — clarity and initiative are. The real constraint is knowing what to build and taking the first step. When everyone has access to GenAI tools, the differentiator becomes who can turn ambiguity into action — who can define a product idea crisply enough that it’s worth prototyping, testing, and iterating. The ability to quickly move from concept to testable product is now in everyone’s hands, not just engineering’s.
At Talkdesk, a PM with no recent coding experience used AI tools to fully prototype a complex product feature — applying design guides, wiring user flows, and demoing it to customers. This wasn’t a weekend hack. It was production-ready, which simply would not have been possible until previously, until the engineering team created space in the backlog.
Raji Subramanian, a stealth Startup CEO and former CTO of a public company, framed it as a move from thinking about the software development lifecycle to thinking about product development lifecycle. In a GenAI-native world, success isn’t just about writing great code — it’s about choosing what to build in the first place.
That means rethinking how problems are framed, how opportunities are sized, and how solutions are validated — all before a single line of production code is written. As she put it: “That is very different. It’s about how do you actually change the what as well as the how.”
This opens up massive opportunities for startups to experiment earlier and faster. But it also places new expectations on team members. If anyone can prototype, anyone can build. Suddenly, a PM, designer, or junior dev might be your most effective accelerator.
The tactical takeaway: Rethink prototyping and feature definition processes. This means empowering every team member — not just product and engineering leads — to define, prototype, and validate ideas. Create space for anyone to prototype.
Tactical Velocity Beats Theoretical Rigor
Velocity has always mattered in startups — but GenAI changes how teams define it, measure it, and operationalize it. With AI tools dramatically reducing the time it takes to go from idea to implementation, the pressure isn’t just to move fast — it’s to know where to aim.
That means founders and leaders must shift from tracking traditional output metrics to evaluating impact velocity: How quickly can your team validate ideas, incorporate feedback, and adjust course?
Munil Shah at Talkdesk, for example, tracks velocity not in lines of code, but in time-to-launch and the percentage of support issues handled autonomously. Meanwhile, Saju Pillai at Kong maintains strict review standards for anything that reaches production, but encourages experimentation earlier in the cycle.
The tactical takeaway: Bias to act. Let a thousand experiments bloom — but know which ones deserve polish and promotion.
Looking Forward
The old SDLC isn’t being optimized — it’s being replaced.
At Madrona, we’re not just talking about this future. We’re investing in it. We’re actively backing startups building and applying GenAI, including the infrastructure it runs on and the tools it enables — from copilots to product agents and beyond. We bring decades of company-building experience and a growing network of technical founders working through these same shifts in real-time.
We believe the next great companies won’t just use GenAI. They’ll be built around it. And we’re here to help you do just that.