How Edmunds GovTech Reclaimed 550+ Dev Days

When Bonsai Labs came in, AI adoption across Edmunds GovTech's product and engineering organization was low. AI licenses acquired through the company's enterprise Microsoft agreement had never been operationalized: no program, no shared tooling, and no proof point the teams trusted. For an organization carrying live delivery commitments to thousands of public-sector customers, that caution was warranted. At the same time, Edmunds GovTech needed to accelerate a generational modernization, its next-generation Ethos web platform, without compromising the stability its customers depend on.

Bonsai Labs embedded with Edmunds GovTech in a forward-deployed model and built AI capability into how the organization works, function by function, from product management to engineering and QA. The centerpiece was a durable AI coding harness designed around Claude Code: documented, extensible, and easy to replicate across every codebase, with teams empowered to add their own skills. A custom toolkit carried a written product vision through specifications, requirements, and design in minutes. The engagement was designed so the capability stayed in-house after Bonsai left.

Bonsai Labs worked in a forward-deployed model, embedding with Edmunds GovTech's teams rather than handing over a deck. They started with product management, demonstrating a custom toolkit that carried a written vision through specifications, requirements, and design in minutes, work that normally took weeks. Winning that team created the first internal believers. Leadership followed by example: the CPTO, originally a coder, turned the Claude Code harness loose on one of the company's applications after a single pairing session and had a click-through prototype hours later. From the outset the engagement was built for durability. Bonsai left behind an AI coding harness designed around Claude Code, documented, extensible, and easy to replicate across every codebase, and taught teams to extend it with their own skills. The work moved function by function, from product management into engineering and quality assurance, pairing hands-on tooling with change management so adoption held after the engagement ended. Teams now extend the harness themselves and operate with an assumption of continuous change.

Private equity-backed B2B software companies modernizing legacy platforms, and product and engineering organizations whose AI adoption has stalled at experimentation and who want a durable, owned capability rather than a purchased tool.