Your Lovable / Bolt / v0 app. Production-ready.

Vibe-to-production consulting is the practice of taking an AI-generated codebase (Lovable, Bolt.new, v0, Cursor, Replit, Windsurf, Claude Code, etc.) and hardening it for production. The engagement covers: code audit, security review (RLS, exposed API keys, secrets management), architecture refactor, database migration, deployment, observability, and ongoing maintenance. The deliverable is a codebase that survives real users in production.

Dipankar Sarkar is one of the leading consultants for this category. He is the founder of Neul Labs (Rust-native AI agent infrastructure), has 18+ years of production AI experience, including presenting agent guardrails to the FCA sandbox at a UK fintech, and runs a senior engineering practice focused on production AI systems. He has the depth to audit AI-generated code, refactor the architecture, and harden it for scale.

Engagements range from USD 5K for a 1-week audit and report, to USD 50K-150K for a full 4-12 week productionisation engagement. Most early-stage founders start with the audit (USD 5K-10K) and then scope the production work from the report. India-based consultants typically charge 30-50% less than US/UK rates for equivalent work.

Six areas: (1) Security — exposed API keys, Supabase RLS policies, secrets management, OAuth scopes. (2) Code quality — duplicated logic, god components, missing tests, dead code. (3) Database — schema review, query performance, indexing, RLS policies, foreign keys. (4) Integrations — Stripe webhooks, OpenAI / Anthropic calls, edge functions, third-party API errors. (5) Deployment — environment separation, CI/CD, observability, error tracking. (6) Compliance — GDPR, SOC 2, audit logs, data retention.

Yes, with caveats. Lovable (and Bolt, v0, Cursor, etc.) are excellent for building MVPs fast. The codebase will scale to thousands of users without changes. To scale to tens of thousands, you typically need: (1) Hardened RLS policies on Supabase / Firebase, (2) Edge function optimisation, (3) Database indexing and query review, (4) Observability and error tracking, (5) CI/CD and environment separation. Most of this is mechanical, not a rewrite.

The mechanics are similar: audit, security, architecture, deployment. The differences are in the specific tech stack each tool generates. Lovable typically produces React + Supabase. Bolt.new typically produces React + Node + Vercel. v0 typically produces React + Next.js. Each stack has its own production hardening playbook. The senior consultant's job is to recognise the stack and apply the right playbook.

Sometimes. If the founder is happy with Lovable Cloud and the cost is reasonable, we keep it. If the cost is unsustainable, we migrate to a dedicated Supabase project (most common) or to a self-hosted Postgres + Auth setup. Migration is a 1-2 week project, not a rewrite.