Skip to content

Roadmap

This page is our honesty layer: what already works carries its real maturity label, and what is coming is announced without selling it as done. Today’s conformance is ex-ante (before production); what is coming extends the method to the operational lifecycle.


KAG — risk identification assistant (future)

Section titled “KAG — risk identification assistant (future)”

🧪 Experimental — The KAG (identification assistant) will propose risks NOT yet declared from the system context (Croissant metadata, historical metrics). Today it is NOT available: the deterministic mock was removed and `froga assess` is an honest stub; the evaluation is declared by hand. The real API service is not productised. That is why the subcommand that invokes it is hidden from the public help.

The KAG (Knowledge Assessment Graph) will assist the risk identification loop (ISO 23894 §6.4.2): it will look at the system context and propose risks not yet in the AssuranceProgram, so the human can curate them. It will be advisory, not blocking. Today identification is done by the human, by hand, in froga.yaml; the real service (Python over an API) is future work.


🚧 Planned — Today's conformance is ex-ante. Post-market monitoring and production monitoring are planned.

The AI Act requires monitoring of the system once in operation (Art. 72). Today Venturalítica attests conformance before deployment; the production proxy and runtime monitoring that close that loop are planned.


🚧 Planned — Today human oversight is declared ex-ante in the AssuranceProgram. Instrumenting it at operation time (HITL at runtime) is future work.

Human oversight (human-in-the-loop) is declared today ex-ante. Instrumenting it at runtime — recording the real human interventions on the system’s decisions in operation — is a future line.


🟡 Partial — Linux, macOS (x86_64 + aarch64) and Windows are all runnable today from the releases bucket. The only pending piece is the vanity URL get.venturalitica.ai (Edge Services TLS, activation pending).

The froga binaries for Linux, macOS (x86_64 and aarch64) and Windows are already served from the releases bucket with valid TLS. The canonical installer at get.venturalitica.ai will be available once Edge Services is activated. In the meantime, Course 1 shows the FROGA_INSTALL_BASE one-liner that works today.


Missing something?

This roadmap moves with what you need. Tell us what is missing or what you would prioritise: hola@venturalitica.ai.