Mentior runs assistants, autonomous agents, and document knowledge bases entirely inside your own infrastructure. Local models, zero external calls — your data simply never goes anywhere.
Early access for teams running AI on-prem. No spam, just the launch.
Build, connect, and observe AI systems without a single byte leaving your perimeter.
The whole system operates inside the customer's own network with no external communication. No telemetry, no API round-trips, no vendor in the data path — verifiable air-gapped operation that satisfies the strictest compliance and data-residency requirements.
Ingest PDFs, text, and markdown into local vector search. Assistants answer from your documents, with context grounded in your sources.
Define a system prompt, pick a default model, and link knowledge bases. Sessions can override the model per chat for flexibility.
Drag-and-drop canvas to chain assistants, agents, knowledge, and tools. Run them with live, step-by-step node status streamed in real time.
Token-by-token responses over WebSocket, with tool-execution logs and RAG context surfaced inline as the answer is generated.
Every prompt, completion, tool call, and latency is traced and stored locally. See exactly what your models did and why — full auditability without shipping logs to a third party.
Connect agents to filesystem, web search, and database tools through a pluggable registry — safe, scoped, and easy to extend.
One stack runs everything — database, vector store, local model inference, and observability — inside your own infrastructure.
Upload documents into knowledge bases. They're chunked, embedded, and indexed for fast local retrieval.
Wire prompts, models, knowledge, and tools together — visually for workflows, or directly for chat.
Teams chat and run workflows while every step is traced locally for full visibility and control.
Join the waitlist for early access to the self-hosted Mentior platform.