Bolt.new Not Working? Check Whether the Project Exceeded the Prototype Layer

Bolt can produce a fast prototype, but failures often appear when the project requires stable state management, backend persistence, auth, database schema, deployment, or external integrations.

Initial verdict

Short answer

high risk

If the prototype looks good but breaks once you add state, persistence, auth, or integrations, the project has likely moved beyond the easy AI-fixable layer.

Tool issue vs build failure

If the tool opens but the project keeps breaking, the failure is often inside the build, not the AI service.

Tool outage

The service is down, rate limited, or failing to generate at all. Check status, account state, and whether the model can answer a simple prompt.

Account or browser issue

The tool opens inconsistently because of login, cookies, extensions, local browser state, or network filtering rather than your codebase.

Prompt or context issue

AI has partial information, weak task boundaries, or too much chat history, so each fix is local and unstable.

Project complexity issue

The request crossed from simple generation into state coordination, auth, payment, deployment, or long-lived maintenance.

Architecture or deployment failure

The code can compile or partially work, but the real failure sits in system boundaries, data flow, environment mismatch, or production setup.

Free self-check

Quick check: is this still safe for AI to fix?

  • 1.Has AI already failed more than twice?
  • 2.Does the problem involve auth, database, payment, permissions, or deployment?
  • 3.Did one AI fix break another part of the app?

If any two are true, this is probably not a simple prompt issue.

Get a 1-page failure diagnosis

Smallest safe next steps

Pick the smallest next action that reduces risk before the AI changes more of the project.

Continue prompting

Narrow the problem

Rebuild context

Stop and redesign

Ask for human architecture review

Related links

Diagnosis

Before AI rewrites more files

If AI has already failed multiple times, the next prompt may make the project worse. A 1-page diagnosis identifies the likely failure layer, why AI keeps failing, what AI can still fix, what AI should not touch, and the smallest safe next step.

Early review: $29 · 1-page diagnosis · no full repo required

Find whether your Bolt project is still AI-fixable