KAONOMI: one conversation, a whole team of AI agents
The multi-agent chatbot and unified interface for every Kaonomi.net solution, powered by your local LLMs and open to frontier AI.
Most chatbots send your question to a single model and hope for the best. KAONOMI works differently. Behind every message, a team of specialised AI agents collaborates: one checks what comes in, another understands the entities and the intent, another picks the most suitable model, another shapes the answer, and a last one verifies what goes out. The result: answers that are more relevant, safer and better adapted to you.
A real multi-agent brain
- A pipeline of 13 agents for every message: guardrails, smart cache, entity extraction, symbolic analysis, model selection, emotion engine, forecasting, context update…
- Cognitive agents running in the background: Reflex reacts in milliseconds, Vision and Metacognition analyse continuously, and Dreamer consolidates context overnight, just as sleep consolidates memory.
- Guardrails on both sides of the conversation.
Your models, your data
- Local LLMs first: your conversations stay on your own infrastructure.
- The right model for each task, chosen by you or by the model-selection agent.
- Open to frontier AI when a task needs it.
- Full transparency: every LLM call is measured in a live tokens dashboard.
One interface for everything
- Natural conversation at a comfortable reading pace, code window, export to JSON or Word.
- Knowledge tools: ontology editor, symbolic patterns with an AI regex assistant, cache, prompt library.
- Built-in supervision: logs and tokens dashboards, system monitor (CPU, GPU, VRAM, temperature…).
- Side panel docked to the screen, system-tray menu, news ticker, over a dozen themes, simple
/commands.
Built for professionals
Open REST API with Swagger, every dialog and measure stored in Elasticsearch and Kibana, runs on Windows and macOS.
Distributed AI and meta AI
At Kaonomi.net we believe the next leap in AI will not come from one bigger model, but from many intelligences working together, and from systems able to observe and improve their own reasoning.
Discover Kaonomi.net · Explore M-A-R-I on GitHub
