Over the past year, we've worked with companies building AI into their products, operations, and internal tools.
Through the Builders Collective, a community we created for service company founders, we recently hosted a roundtable where founders compared notes across hundreds of AI projects.
Three things kept coming up:
Proof of concepts and MVPs that used to take quarters can now take weeks. That changes what's worth building for in the first place.
Problems that were previously too niche to justify may now be practical to build. Things like highly specific internal tools, system-to-system workflows, and nuanced processes. The cost of building something custom has dropped. The value of a precise fit has gone up.
The market data tracks with what we're seeing on the ground. In Retool's 2026 Build vs. Buy report, 35% of teams said they'd already replaced at least one SaaS tool with something custom, and 78% expect to build more of their own tools next year.
We saw this firsthand. We built an internal CRM connected directly to our staffing allocation tool, so resourcing updates automatically as deals move through the pipeline. No off-the-shelf product supported the exact workflow we needed. A few years ago it would have been a major project. Instead, one engineer built an MVP in just over two weeks.
Others in the room had similar stories:
That pattern is showing up everywhere. Even when companies do buy software, it often only gets them 80% of the way there. The last mile — connecting the product to your specific users, customers, and workflows — still needs to be built.
When build cycles compress, getting the problem and scope right upfront determines the speed and quality of everything that follows.
Teams can now ship the wrong thing faster than ever. A team that misframes the problem or scopes too broadly will find out sooner, but not before spending the effort.
Discovery and scoping have always mattered. What's changed is the relative importance. Now that building is fast, scoping is by far the most important human step in the process. The teams moving fast aren't always the best funded or resourced. They're the ones making better decisions at the start.
We've seen this with our own work. Some of the most successful projects we've been part of had one thing in common: a well-defined problem. Clarity on what to solve, and what not to solve, made everything downstream faster.
In the AI era, speed is a function of both knowing what to build and having the tools to build it radically faster than before.
For a long time, the software playbook was familiar. First-mover advantage, proprietary technology, network effects. These were the moats that protected companies and gave them time to iterate.
AI has compressed those timelines dramatically. Companies scale faster, competitors catch up faster, and the cost of moving slowly compounds.
Earlier this year, that pressure became visible at the market level. A Claude update triggered a selloff across publicly traded SaaS — a moment some have started calling the "SaaSpocalypse". Sam Altman described the broader dynamic as software's fast fashion era: cheap and fast enough that anyone can spin up a tool over a weekend.
When anyone can build a prototype in a weekend, the advantage stops being who got there first or who has the better technology. It becomes who learns the fastest from real customers and turns that learning into the next version.
The new moat is execution speed. A couple of examples:
The winners are no longer just those with the best ideas. They're the fastest learners and executors. And speed without scoping discipline is just churn at a faster clip. Across the projects we've been part of, the companies that moved fastest were almost always the ones that took the time to scope carefully at the start and prioritize the right features. That's the quiet cost of the new economics: the bottleneck moves upstream, to whether you scoped the right problem in the first place.
AI hasn't just changed what's possible to build. It's changed what it takes to build well.
Thanks to those who joined the first Builders Collective roundtable: Philip Clements Samuelraj (Techjays), Harit Patel (Lodestone), Jasmeet Kanwar (Matchpoint), Jimmy Bijlani (AI Momentum Partners), Ajay, John Koshy (PeakIT), and Thanujan Ratnarajah (Verdant Labs).
