We’re seeing a new objection pattern emerge on compliance calls that wasn’t common two years ago. Prospects are building their own AI infrastructure instead of buying vendors like us.
Last week, we connected with a CTO at a fintech company who walked us through their custom internal gateway. They’d built token optimization, custom compliance routing, and audit trails entirely in-house using Claude and some open source tooling. The conversation lasted 17 minutes. His conclusion: they don’t need us. They need customers. He said it plainly: “Our bottleneck is sales, not infrastructure.”
This is happening more often. In our last call blitz, we saw three deals stall or close when prospects revealed they had similar builds underway. One told us they’d defer any vendor decision for six months minimum, until their internal solution was production-hardened. Another was already two months into development. The pattern is unmistakable.
Why it’s happening now is straightforward. Claude and open-source LLM APIs made building internal AI systems genuinely feasible for engineering teams with 10-50 people. A CTO can prototype a compliance gateway in two sprints. Add some prompt engineering, a vector database for audit trails, and token accounting, and you’ve got something that looks like a vendor product. The build-versus-buy calculation shifts when your team is already trained on LLMs and can build in a month what would have taken six before.
The threat isn’t that these DIY systems are better than ours. They’re not. They’re missing things: scalability, regulatory hardening, support, cross-application compatibility. What they do have is ownership and cost control. A prospect who’s already spent engineering time and budget defending a build to leadership won’t switch vendors lightly. They’re locked in by sunk cost and organizational momentum.
For fintech and compliance-heavy verticals, this matters immediately. We’re competing not just against other vendors but against a prospect’s engineering roadmap. When a CFO is already allocating budget for internal development, the vendor conversation moves to a “maybe later” timeline.
The secondary effect is subtle but real. Prospects who build in-house become more sophisticated buyers if they eventually come back to vendors. They know what good looks like. They know what corners they’re cutting. They ask harder questions. They’re less likely to buy based on feature lists and more likely to audit integration complexity and operational burden. These are better sales conversations, but they’re longer and have lower conversion rates.
What we’re seeing is a structural shift in how enterprises approach compliance automation. The barrier to entry for building your own AI solution just dropped below the bar where buying made obvious sense. This isn’t a phase. It’s the new normal.
The tactical play is straightforward: get in earlier, before engineering commits to a build. Position us as the proof of concept or the fast path to production while they validate the business case. If they’re already three months into a build, positioning matters less than honest conversations about timeline and scope.
The bigger picture is that compliance automation is no longer a vendor-only game. We’re competing in a world where more prospects have the skills and capital to build. Some will succeed. Most will stall. But all of them will cost us six to twelve months of sales cycle they wouldn’t have six months ago.