Migration snapshot
ActivePlatform
a Retool
Retool reached ~$120M ARR by Oct 2025 (+39% YoY, Sacra est.), driven by AI AppGen/Agents (public beta Oct 2025). Feb 2026 community thread confirmed self-hosting is now effectively Enterprise-tier only, reversing the 2021 open self-host position (community.retool.com/t/self-hosted-docs-now-state-enterprise-only/64586). Business tier is $65/standard user/mo.
Typical timeline
8–12 weeks
Typical cost
$13K–$25K (agency, fixed)
Why teams leave a Retool
Retool is excellent at what it does — the migration trigger is almost always economic or driven by lock-in as headcount grows.
Per-seat cost spiral
Business tier at $65/standard user/mo. At 20+ users this exceeds the amortized cost of maintaining an in-house admin framework (Sacra 2025). Model your 12-month projected headcount before deciding.
Self-host now Enterprise-gated
A Feb 2026 community forum thread confirmed self-hosting is no longer available on Team/Business tiers — only Enterprise. This reverses Retool's 2021 open self-host position and changes the cost model for teams that relied on it.
No code eject
Apps export as JSON but run only on Retool's runtime (cloud or self-hosted Docker/K8s). Exit means a full rebuild regardless of what you export — the JSON is a config spec, not runnable code.
Feature gating
SSO, audit logs, granular RBAC, and Git/source control are all gated to Business/Enterprise tiers. Free tier is capped at 5 users. As your compliance requirements grow, so does the tier cost.
AI lock-in acceleration
AppGen/Agents (launched Apr–May 2025) tie workflows deeper into the proprietary runtime. The longer you build with AI-generated Retool apps, the costlier the eventual exit becomes.
What can you actually take with you?
Your source data is always in your own databases — Retool is a query layer, not a data store. The app itself (UI, logic, workflows) has no source-code eject path.
| Asset | Can you export it? | How | Notes |
|---|---|---|---|
| Data | Yes | Retool does not store your source data — queries run against your own databases (Postgres, MySQL, REST, etc.); your data is already yours and always has been | Retool Database (if used) is Postgres via Neon — export via standard pg_dump |
| Code | No | Apps export as JSON config only via Retool Settings → Export | JSON runs only in the Retool runtime (cloud or self-hosted container); it is not runnable standalone code |
| Design/UI | No | Component layout is Retool-proprietary; no export to HTML or React | Screenshot every page and state for use as a parity specification during rebuild |
| Logic/Workflows | Partial | SQL and JS queries are readable inside the exported JSON; workflow and AI Agent definitions are also present in JSON | Logic is tied to the Retool runtime — readable as a spec but must be rewritten as server actions or Edge Functions |
| Users & Auth | Partial | SSO users (SAML/OIDC) are managed by your identity provider and are provider-side portable; built-in Retool users require a forced password reset | Password hash export is not documented; SSO provider config (Okta, Azure AD) is portable to the new platform |
| Scheduled Workflows | Partial | Trigger cadences are visible in JSON export and in the Retool Workflows UI | Must be rebuilt as Supabase Edge Functions or Vercel Cron; no portable format |
Swipe the table sideways to see the full breakdown.
Where each piece moves in code
The target stack is Next.js (App Router) + Supabase/PostgreSQL — a maintained, self-hostable alternative that eliminates per-seat runtime costs.
a Retool
Retool queries (SQL/JS over resources)
In code
Next.js Server Actions + Supabase/Drizzle query layer
Extract SQL from JSON export; refactor into typed service functions
a Retool
Retool resources (DB connections, REST APIs)
In code
Supabase + environment-variable–secured Next.js Route Handlers
Connection strings move to .env / Vercel environment variables
a Retool
Retool components (tables, forms, buttons)
In code
shadcn/ui + Radix UI React components
Use exported JSON screenshots as parity spec; rebuild component by component
a Retool
Retool workflows (scheduled + event-driven)
In code
Supabase Edge Functions or Vercel Cron
Document every trigger cadence from the Retool Workflows UI before canceling
a Retool
Retool AI Agents / AppGen
In code
Custom AI SDK integration (Anthropic/OpenAI) via Next.js Route Handlers
AI Agent definitions are in JSON export but must be reimplemented as server-side AI calls
a Retool
Retool RBAC / user groups
In code
Supabase RLS + role column in users table
Map Retool permission groups to Supabase RLS policies before cutover
a Retool
Retool Audit Logs
In code
Supabase table with server-side write on every mutation
Enterprise-gated in Retool; implement as a standard Supabase insert in Next.js Server Actions
a Retool
Self-hosted Retool (Docker/K8s)
In code
Vercel + Supabase cloud, or self-hosted Next.js on Railway/Fly.io
Eliminates the Enterprise-tier self-host gate and Docker ops overhead
The migration roadmap
Plan 8–12 weeks. The largest time sink is auditing SQL/JS logic embedded in component props — allocate this time before writing a line of replacement code.
Extraction & Audit
Week 1–2- Export all app JSON via Retool Settings → Export
- Inventory every resource: document name, type, connection string, auth method
- Screenshot every page and every query output for parity testing
- Audit which users are on built-in auth vs SSO
- Document all scheduled workflow triggers and their cadence
Watch out: SQL/JS query logic embedded in component props is easy to miss — do a systematic export review, not a spot check
Foundation Setup
Week 2–3- Scaffold Next.js (App Router) + Supabase project
- Migrate all resource connection strings to environment variables
- Set up Supabase RLS schema and role column mirroring Retool user groups
- Configure SSO provider redirect URLs if applicable
Query & Logic Migration
Week 3–6- Extract SQL queries from JSON export; refactor into typed Drizzle/Supabase service functions
- Rebuild Retool Workflows as Supabase Edge Functions or Vercel Cron
- Port AI Agent logic to Anthropic/OpenAI SDK Route Handlers
- Implement audit log table with server-side writes
Watch out: Complex apps with 50+ queries may need a dedicated 2-week query-audit sprint
UI Rebuild & Feature Parity
Week 5–9- Rebuild pages with shadcn/ui components using screenshots as spec
- Implement data tables, forms, and modals page by page
- Validate each page against the screenshot spec before moving on
- Build admin auth guard in Next.js middleware
Parallel Run & Cutover
Week 9–12- Run Retool and new app in parallel for at least 2 weeks
- Switch teams page-by-page, not all at once
- Trigger forced password reset for built-in Retool users on new platform
- Update SSO provider redirect URLs to new domain
- Decommission Retool subscription after all teams confirm parity
Three ways to migrate — honestly
Every path has a real trade-off. Here is what each costs, how long it takes, and where it bites.
DIY (with AI tools)
$0–500 + your time
4–6 months part-time
Fits
Founders or developers who can read the Retool JSON export and translate it to React/Next.js; works best for simple apps with 3–5 pages and limited query complexity
Risks
Query logic embedded in component props is easy to miss; AI tools (Cursor, Copilot) accelerate UI but don't auto-migrate business logic; risk of feature gaps discovered post-cutover
Freelancer
$5K–12K
6–10 weeks
Fits
Apps with moderate complexity (10–20 pages, SQL queries, basic workflows); freelancer needs solid Next.js + Supabase experience and ability to read Retool JSON exports
Risks
Single-point-of-failure; vetting is on you; no fixed-price guarantee; complex Retool Workflow/Agent migrations may be quoted separately
Agency (RapidDev)
Done-for-you$13K–25K fixed price
6–10 weeks
Fits
Internal apps with 10–40 pages, complex SQL queries, workflows, SSO, and audit log requirements; teams that need guaranteed parity and a managed cutover
Risks
Highest upfront investment; right choice when per-seat cost savings over 12–18 months cover the project cost — model this before booking a free scoping call at rapidevelopers.com
The real risks — and how to defuse them
Query logic loss
Mitigation: Retool SQL/JS queries embedded in component props are easy to miss in a surface-level JSON review. Run a systematic export audit, component by component, before writing replacement code.
Auth disruption
Mitigation: Built-in Retool users will need a forced password reset on the new platform. SSO users need redirect URL updates at the identity provider. Audit which user type each person is before cutover.
Feature-parity trap
Mitigation: Retool has 100+ built-in connectors that take weeks to replicate. Prioritize the 20% of connectors used 80% of the time; build the rest on demand post-launch rather than blocking cutover.
Workflow and Agent rebuild
Mitigation: Retool Workflows and AI Agents have no portable format — each must be rebuilt as an Edge Function or Server Action. Document every trigger, action, and schedule from the Workflows UI before starting.
Self-hosted Retool 4.0 migration debt
Mitigation: If already self-hosted, upgrading to Retool 4.0 requires Docker Compose/ECS → Kubernetes/Helm migration before exit. Factor this pre-migration cost into your timeline and budget.
Should you actually migrate?
Migrating is a real project. Sometimes staying is the right call — here is the honest split.
Stay if
- Your team is 10 users or fewer and per-seat cost is under the amortized cost of 1 engineer's time to maintain a custom stack
- You need rapid internal tooling iteration and have no custom UI requirements — Retool's 100+ connectors and drag-drop builder beats months of engineering for simple ops tools
- You're on Enterprise with self-hosting, SSO, and audit logs fully configured and compliance requires an established vendor
Migrate if
- Standard-user headcount × $65/mo is approaching or exceeding the cost of 1–2 engineers at 12-month projected headcount
- You need customer-facing polish or custom UI beyond what Retool's component set allows
- Self-host re-gating to Enterprise tier has changed your cost model and you can't justify the Enterprise contract
Our honest verdict
Retool is excellent for internal tooling at small scale. The migration trigger is almost always economic — model per-user cost at 12-month projected headcount before making the call.
Do this today: pre-migration checklist
Whatever path you choose, protect yourself first. Work through this before you touch a line of code.
Export all app JSON via Retool Settings → Export
This is your migration spec — every query, component, and workflow lives in this file
Inventory every resource: document name, type, connection string, and auth method
Resources are the connective tissue of your Retool apps; missing one causes silent failures in the rebuild
Screenshot every page and every query output
Screenshots serve as your parity test spec when there is no source code to diff against
Audit which users are on built-in auth vs SSO
SSO users migrate cleanly; built-in users need a forced password reset flow planned in advance
Export Retool Database data if used (pg_dump from the Neon Postgres instance)
Retool Database is the one case where data lives in Retool's infrastructure rather than your own
Document all scheduled workflow triggers and their cadence
Retool Workflows have no portable format; the cadence and logic must be reconstructed from documentation
Validate that your data sources are accessible outside Retool before writing a line of code
Confirms that migration is unblocked — occasionally connection strings have hardcoded Retool-network assumptions
Frequently asked questions
Can I export my Retool app as source code?
No. Retool apps export as JSON configuration files via Settings → Export. This JSON is readable and useful as a migration spec, but it runs only inside the Retool runtime (cloud or self-hosted Docker). There is no source-code eject — exit means a rebuild.
What data can I take with me when leaving Retool?
All of it. Retool is a query layer — it connects to your own databases (Postgres, MySQL, REST APIs) and does not store your source data. If you used Retool Database, that is Postgres via Neon and exports via standard pg_dump. Your data has always been yours.
How long does a Retool migration take?
Typically 8–12 weeks for a full rebuild. Simple apps (3–5 pages, basic SQL queries) can be done in 6–8 weeks. Complex apps with 20+ pages, custom workflows, AI Agents, and SSO requirements can reach 12 weeks. The largest time sink is auditing SQL/JS logic embedded in Retool component props.
What happens to my users and their passwords when I migrate?
SSO users (SAML/OIDC via Okta, Azure AD, etc.) are managed by your identity provider — they migrate cleanly with a redirect URL update at the IdP. Users on Retool's built-in email/password auth will need a forced password reset on the new platform; password hash export is not documented.
Is Retool self-hosting still available?
As of Feb 2026, self-hosting is effectively Enterprise-tier only. A community forum thread (community.retool.com/t/self-hosted-docs-now-state-enterprise-only/64586) confirmed this change, which reverses Retool's earlier open self-host position. If you relied on self-hosting on a Team or Business tier plan, your cost model has changed.
What about Retool's AI Agents and AppGen features?
AI Agent definitions are included in the JSON export and are readable as a spec. However, the runtime logic is tied to Retool's proprietary infrastructure and must be rebuilt as server-side AI SDK integrations (Anthropic or OpenAI via Next.js Route Handlers). The longer you build with AppGen, the more runtime dependencies accumulate.
How much does a Retool migration cost?
DIY with AI tools: $0–500 plus your time (4–6 months part-time, works for simple apps). Freelancer: $5K–12K over 6–10 weeks. Agency with a fixed-price guarantee: $13K–25K over 6–10 weeks. RapidDev offers fixed-price Retool migrations with a free scoping call at rapidevelopers.com — book one before committing to any approach.
Is it worth migrating if Retool is still active and growing?
It depends entirely on economics. Run this calculation: (current users + 12-month projected new users) × $65/mo × 12 months vs. a one-time migration cost. Many teams find the crossover happens between 15–25 users. If you're well below that threshold, Retool's speed advantage likely outweighs the migration cost.
We migrate no-code apps to production code
- Fixed price — $13K–$25K (agency, fixed)
- No data loss, no downtime
- You own 100% of the code
30-min call. Quote within 48 hours.