Your data stays where you deployed the platform.
Self-hosted keeps every byte inside your VPC. Hosted lets you pin residency to EU or US. Either way, your data plane and our billing plane never mix.
How it works
From a paragraph of intent to a running application.
Three steps. Deploy the AnySaaS platform — hosted by us or on your own infrastructure — then describe your app to the assistant and it builds it in place. One platform, unlimited apps, unlimited users.
Deploy/Describe/Build
The pipeline
Three steps that stay three steps.
We resist adding a fourth. Every extra hop is a place for enterprise IT to lose an afternoon.
01Deploy
Run the platform where you want it.
AnySaaS ships as one platform — one Docker image, identical for every customer. Let us host it as a typical SaaS, or self-host it on your own infrastructure. Same features either way.
- Hosted by AnySaaSSign up, pick a region, you’re in. Managed updates, 99.9% SLA. No infrastructure work.
- Self-hostedOne command, one host. AWS, Azure, GCP, on-prem, private cloud, air-gapped — anywhere Docker runs.
- One image, everyoneThe self-hosted image is the same one we run in the cloud. No fork, no bespoke build.
- Zero manual securityAuth, RBAC, encrypted database, audit hooks: all wired in before you start building.
$ docker compose pull
$ docker compose up -d
anysaas-platform → healthy
postgres → ready
open → https://anysaas.internal.acme.co02Describe
Tell the assistant what you actually need.
Inside the running platform, skip the wireframe and the six-week discovery. Type — or paste — the shape of the tool: the entities, the workflows, who can do what. The assistant asks back only when the answer would change the schema.
- Plain-English intentNo prompt engineering. "Internal CRM with a deal pipeline and manager approvals" is a valid input.
- Grounded questionsIt asks about roles, retention, and edge cases — never about colours or copy.
- Design-system awareEvery screen it proposes uses the same audited UI patterns. No bespoke, no drift.
- Bring your own modelPoint the assistant at OpenAI, Anthropic, a local model, or use ours. Configurable per deployment.
› Build an internal CRM for our sales team with contact management, a deal
› pipeline, and role-based access for managers and reps. Managers can approve
› discounts above 15%. Reps can only see their own accounts.
◇ planned entities → Contact, Account, Deal, Approval
◇ planned roles → rep, manager, admin
◇ planned screens → 7
◇ open questions → 203Build
The platform builds the app in place.
The AI generates the schema, the API, the screens for every role, and the RBAC to enforce them — all inside the running platform. Nothing to package, nothing to redeploy. Every generation is a versioned diff you can accept, edit, or roll back.
- Built inside the platformApps live in the platform — there is no per-app container to ship. Deploy the platform once, build as many apps as you want.
- Standardized patternsAuth, sessions, encryption at rest, audit hooks are already there. Every generated app inherits them by default.
- Versioned generationsEvery publish is an immutable version. Diff against the previous one; roll back with one click.
- Portable dataData lives in standard Postgres. First-class export and erasure endpoints — your data is yours to take with you.
◇ generating schema → ok
◇ generating screens → 7 built
◇ wiring rbac → ok
◇ version → [email protected]
open → https://anysaas.internal.acme.co/apps/crmGuarantees
Three commitments the model can’t break.
The pipeline is opinionated on purpose. These are the invariants we do not let the assistant negotiate away, no matter how the prompt is worded.
One design system, audited once.
Every screen we generate uses the same audited patterns. Security fixes ship with the platform and apply to every app you have built inside it.
Standard Postgres, first-class export.
Your data lives in a Postgres you can back up, query, and export. Erasure and export endpoints are built in — no proprietary format, no lock-in on your data.
See it end-to-end in three minutes.
A short walkthrough: deploy the platform, describe the app, watch the assistant build it in place. Or skip the video and read the deployment guide.