We built Buildly because Q3 2022 was entirely CRUD endpoints.
Not a slow quarter. An entire sprint cycle — migration scripts, pagination endpoints, field additions. Senior engineers writing code that didn't require senior engineers. We were those engineers.
One bad quarter. One prototype. Three years of building.
Camille Fontaine and Marcus Chen were engineering leads at Ardent Payments — a fintech processing platform that was growing fast. In Q3 2022, their team spent an entire quarter on data migration endpoints. Feature velocity was essentially zero.
The work wasn't hard. It was just relentlessly repetitive. Same patterns, same module structure, same conventions — just a different table name or field set each time. They started asking: why are senior engineers doing this?
After that sprint, Camille spent a weekend building a prototype that could read a Jira ticket and generate code matching Ardent's module structure and naming conventions — not generic output, code a reviewer would recognize as belonging to their repo. The first real ticket it handled needed no revision. They founded Buildly in San Francisco in 2023 to turn that weekend proof-of-concept into a production agent.
Angel-backed in 2025. We're working with 15+ engineering teams — mostly at fintech platforms, developer tools companies, and mid-market SaaS. They all had the same frustration: Copilot-style per-seat autocomplete tools helped individual engineers write faster, but didn't reduce the number of boilerplate tickets in the backlog. That's the problem Buildly is built to fix.
Three people. One specific problem.
Previously engineering lead at Ardent Payments, where she spent one too many sprints writing migration scripts. Built the first Buildly prototype over a weekend to prove the concept to herself.
Backend infrastructure engineer with 12 years in distributed systems. Designed the Style Graph engine and the PR-only write architecture. Believes the best code review is reviewing code that already works.
Joined from a developer tools startup where she built static analysis pipelines. Leads agent quality at Buildly — obsesses over false positive rates and precision.
How we think about this.
The agent's job is to give engineers something to review, not to make decisions for them. Every Buildly write ends in a PR. We will never ship a "direct commit" mode. The constraint is the feature.
A false positive doesn't just waste a review cycle — it trains engineers to distrust all agent output. We'd rather return a clarifying question on the ticket than ship a PR that looks right but isn't. Buildly defaults to conservative confidence thresholds. We are not optimizing for volume of PRs opened.
We don't homogenize. Buildly adapts to your team's style, not the other way around. The Style Graph exists precisely so the agent writes code that belongs in your repository — not code that looks like it came from somewhere else.