The goal of the AX Brie delivers is to build a system where agents actually do the work. Before we bring it to clients, we run it inside our own company first — meeting notes, email, and work files accumulate into an internal Knowledge Management System (KMS), and agents handle the work themselves on top of that work context.
Instead of a person opening a tool to work, the agent reads the work context, decides for itself, and does the job. People review results rather than operate tools.
Meeting notes, email, and work files are turned into a KMS by the internal knowledge system (Knowclaw), and multiple agents work on top of that work context.
As notes, email, and files build up in the KMS, the work context grows richer and the agents running on top of it make sharper decisions.
They share the same knowledge graph (KMS) and each autonomously handles a different job.
Watches the inbox around the clock, sorts mail on its own and even drafts replies. Work mail is auto-archived to project folders — a person approves before anything is sent.
Uses the meeting notes and work records built up in the KMS to automatically track and summarize the status of company-wide projects.
Agents handle the development of standarda itself — the foundation that powers our client cases. Our own agents build our own tools. Dev automation tools →
Each links through to that system's deck (presentation).
An autonomous (ReAct) agent specialized in writing meeting notes. It observes, records, learns, and finalizes on its own, then feeds the result into the KMS as context.
View deck →A self-growing internal LLM wiki. Through four stages — collect, organize, wikify, ask — it builds the work context shared by all three agents.
View deck →An autonomous agent that watches your inbox for you. A human-supervised pipeline of constant polling → sorting → action leaves only the mail you actually need to see.
View deck →