B2B Moat: Fitting Messy Codebases
Eric Simons, CEO of Bolt, on consumer vs. B2B vibe coding
The real moat in B2B vibe coding is not generating a demo, it is fitting into the messy code and tooling a company already has. Bolt is positioning around that gap. The hard part is ingesting a company’s design system, auth, database choices, and codebase conventions so a PM or product engineer can ship something that engineers keep, instead of rebuilding from scratch. That is why Cursor and Claude Code are the closest analogs, while prototype first tools face weaker retention.
-
Bolt’s landing motion is concrete. Teams plug in their existing codebase, Bolt traverses it, builds documentation, and uses that context to generate code that matches internal components and setup. That makes it useful for product teams working inside a real company stack, not just for greenfield experiments.
-
The market is splitting into two workflows. Lovable, Bolt.new, and similar tools are strong for fast app generation, then users often move into Cursor or another AI IDE for deeper editing. That split matters because the editing layer is closer to the production repo and usually carries higher switching costs.
-
Incumbent pressure is strongest wherever the output is disposable. Figma Make is bundled into existing design budgets for prototype style use cases, while consumer focused builders like Lovable, Replit, and Base44 compete on top of funnel and convenience. The remaining white space is software that plugs into enterprise systems and survives past the first draft.
The category is moving toward a stack where prototype tools, coding agents, and deployment infrastructure blur together. The winners in B2B will be the products that become part of the company’s normal build process, absorb its internal context automatically, and turn non engineer intent into code that ships without a second rewrite.