Product
Three layers, one source of truth.
A structured business model, an engine that carries multi-tenancy, roles, authentication, billing, audit and compliance, and an AI-assisted journey that creates the application and keeps improving it.
The model
Structured, schema-backed, cross-referenced.
- Market snapshot — segments, competitors, positioning — the reasons the product exists, as data.
- Domain plan — the resources, their fields, relationships and lifecycle, in a vocabulary the engine understands directly.
- Requirements and roadmap — what must be true, in what order, traced to the features that satisfy them.
- Compliance posture — which controls the application claims, against which framework generation, with the evidence generated from the implementation.
- Coherence report — dangling references, coverage gaps and contradictions computed from the model, not spotted in review.
The engine
The parts every B2B product rebuilds, built once.
- Resources — declare a schema with decorators and get persistence, validation, HTTP operations, forms, lists, views, search and realtime updates.
- Tenancy and authority — organisations, memberships, roles per scope, fail-closed authorization, platform-operator access that must be granted, never assumed.
- Identity — password, magic link, MFA, social OAuth, enterprise SSO with SAML and OIDC, SCIM.
- Money — billing, subscriptions and invoicing through Stripe.
- Evidence — audit trail with per-organisation SIEM export, subject export, erasure with retention policies.
- Runtime — MongoDB or PostgreSQL, queue workers, long-running minions, email, files, PDF generation, i18n, OpenTelemetry.
- AI — LLM, embeddings and retrieval over resource data, document extraction, agents as flows and actors, MCP and A2A exposure.
The journey
Playbooks, not prompts.
- Playbooks — the methodology — how a business is worked out — expressed as steps a machine can run, gate and replay.
- Coding agents — write the application code into the engine, against specifications derived from the model, in a place computed from what the artifact is.
- Validation — the provider is constrained by the artifact’s schema, the family validator checks the parts agree, the application’s own build is the floor.
- Acceptance — a git commit. Guided mode: a human reviews and commits. Autonomous modes: the system commits, and git keeps every change visible and reversible.
- Companion — a local development companion serves the journey, introspects the codebase and assembles context on demand.
What comes out.
- A running application: backend API, frontend, infrastructure for Docker Compose or Kubernetes.
- Its specifications: the readable, versioned description of every resource, operation and view.
- Generated technical documentation for its users, administrators and integrators.
- A generated marketing website for the application itself.
- Compliance documents derived from what the application actually implements.
Does this fit what you are building?
Wildo is in active development with real applications running on it. Tell us about the product or the internal tool you have in mind.