Generate production systems with AI. Deploy them on your Docker host, your Kubernetes cluster, your on-prem rack, or your air-gapped network. Standard code. Your data. No proprietary runtime between you and your customers.
Hosted AI builders generate code on someone else's runtime. For regulated industries, that's the part that breaks first — at the audit, at the data-residency review, at the renewal.
"Appropriate technical measures" means knowing exactly where personal data lives, who can read it, and how it's encrypted in transit and at rest. With Weezzi self-hosted, your customer data never leaves your VPC.
The EU AI Act requires logging, technical documentation, and human oversight for high-risk applications. Hosted black-box generators don't expose what's needed. Self-hosted Weezzi gives you full code, full logs, full audit.
Standard Java, Python and JavaScript. Sitting in your Git. No proprietary DSL, no opaque runtime, no "export — but you'll need our enterprise plan to actually run it." If we disappear tomorrow, your apps keep running.
Bubble's WU pricing scales with usage. Outsystems charges per developer, per environment, per app. Weezzi self-hosted: pay for the Builder seats that build, run unlimited apps on infrastructure you already own.
Banking, healthcare, defence, public sector. Their procurement teams already have a checkbox for "self-hosted, source-available, no third-party data processor." Weezzi was designed to tick that box on day one.
Acquisition, pivots, pricing changes, sunset announcements — every PaaS customer eventually lives one of these. Source escrow plus standard runtime means your apps survive any scenario, including ours.
From a single Docker container on a developer laptop to fully air-gapped clusters in regulated environments — same generated app, same operating layer, same standard runtime.
The fastest path. One docker compose up brings the whole platform online — Builder, generated app, Postgres, observability — on any developer machine, VPS, or single-server deployment.
Full Helm chart, horizontally scalable services, blue/green rollouts, autoscaling, distributed Postgres, S3-compatible object storage. Deploy on EKS, AKS, GKE, OVH, Hetzner, Scaleway — or your private cluster.
Annual license for fully on-premises deployment in your datacenter. Includes Builder seats, unlimited apps, BYOK AI integrations, SLA tiers, source code escrow. Engineered installation by Weezzi or partner.
Fully offline operation for defence, critical infrastructure, and sovereign workloads. Local LLM inference (Llama, Mistral, Qwen). Signed offline update bundles. Hardened image. No outbound network required.
When other platforms say "self-hosted," they often mean "we let you run our proprietary runtime in your cloud." Weezzi means something stricter: you get the code, the schema, and the binaries — and we lose the ability to make you depend on us.
Java (Spring Boot), Python (FastAPI), JavaScript (modern frontend). No proprietary syntax. Your team can read, debug, fork, and rewrite any part of it.
Apps run on a JVM, a Python interpreter, and a Node process — that's it. No Weezzi sidecar required at runtime. No license check phoning home.
Postgres schema generated by the platform. Versioned migrations live in your repo. Run them with Flyway, Liquibase, or your own CI — Weezzi tooling is helpful, never required.
For Enterprise customers, full source is deposited with a third-party escrow agent. Release triggers — bankruptcy, sustained SLA failure, sunset notice — are spelled out in the MSA, not promised in marketing.
Bring your own keys for OpenAI, Anthropic, Mistral, or self-hosted Llama / Qwen. No Weezzi-mediated inference, no usage logged on our side, no tokens passing through our infrastructure.
Every generated app ships with structured logs, distributed traces, and Prometheus metrics out of the box — emitted in OpenTelemetry format. Pipe them straight into your existing Datadog, Grafana, ELK, or Splunk stack. Dashboards for request rate, error budget, and audit trail come pre-wired.
A short, honest comparison from a self-hosting perspective. We're not the fastest first prototype — we're the one you can take home.
| Capability | Lovable | Bubble | OutSystems | Weezzi |
|---|---|---|---|---|
| Self-hosted production runtime | No | No | Proprietary | Docker · K8s · on-prem · air-gap |
| Generated code is standard, readable | React + Supabase | Visual logic only | Proprietary | Java · Python · JS |
| EU data residency guarantee | Vendor-managed | Vendor-managed | EU regions | Your VPC, your country |
| Air-gapped operation possible | No | No | Limited | Yes · local LLM |
| Source code escrow | No | No | On request | Included · Enterprise |
| Pricing model at scale | Per message / month | Workload Units | $36K+/yr · per-app | €0 self-host · flat license |
| BYOK for AI inference | Limited | Plugin-based | AI Agent · vendor | Any provider · local LLM |
| Lock-in if vendor disappears | App stops | App stops | Migration project | App keeps running |
Most customers self-host the generated apps and continue to use Weezzi Builder in our cloud. Enterprise and regulated buyers can take it one step further: deploy the Builder itself inside their perimeter — so AI generation, the editor, the runtime, and every byte of customer data never leave their network.
For Enterprise on-prem deployments, the Weezzi Builder — the AI generation engine, the visual canvas, the Site Editor — installs alongside your generated applications. Prompts, source code, schemas, and inference all happen on your infrastructure. We can leave the building entirely.
The full Weezzi Builder runs on-prem alongside your generated apps. AI generation, prompts, schemas, source code, and edits never leave your infrastructure. The most sealed deployment we offer.
OIDC and SAML 2.0 via Okta, Microsoft Entra, Keycloak, Authentik, or Ping. RBAC mapped to your existing AD/LDAP groups. SCIM provisioning included.
HashiCorp Vault, AWS KMS, Azure Key Vault, sealed-secrets. AES-256 at rest, TLS 1.3 in transit. Customer-managed keys for every encryption boundary.
Every Builder action, every Site Editor change, every API call recorded with actor, timestamp, payload diff. Streamed to your SIEM. WORM-storage compatible. EU AI Act Article 12 ready.
FIPS 140-3 validated crypto modules. Distroless container images. CIS-benchmarked Kubernetes manifests. Signed releases (cosign) and SBOMs (CycloneDX) for every artifact.
Mission-critical SLA on top of the Enterprise license. Named technical account manager. Quarterly architecture reviews. Penetration test reports on request.
A regulated mid-market deployment, end to end. The full guide ships with the engagement; this is the shape of it.
Two sessions. Network topology, identity provider, secret management, observability stack, backup policy. Output: a deployment plan signed by your security team.
Helm chart applied to your cluster. Postgres, MinIO/S3, Redis, OpenSearch deployed inside your VPC. Zero outbound dependencies beyond container registries you control.
SSO via OIDC (Okta, Entra, Keycloak, Authentik). Secrets via Vault / AWS SM / sealed-secrets. RBAC mapped to your existing groups, not a Weezzi user table.
Open the Builder. Prompt or import. Generated repo lands in your Git, not ours. CI runs your existing pipelines on it.
Final firewall pass. Disable any outbound to Weezzi. Confirm offline operation. Hand over runbooks. You're sovereign.
Book a 30-minute deployment briefing with our solutions team. We'll review your security posture, your compliance scope, and the deployment topology that fits — Docker, Kubernetes, on-prem, or air-gapped.