A Hermes-based AI operator across Slack, GitHub, Linear, Notion, meetings, and the terminal.
ZeroGen Agent turns messy intent into grounded action across real tools.
Inspect files, PRs, project state, and transcripts before answering.
Move important facts out of chat and into durable systems.
Let the agent act where safe; require approval at mutation boundaries.
| Surface | Role | Why it matters |
|---|---|---|
| GitHub | Canonical operating context | Versioned, diffable, reviewable, agent-readable. |
| Linear | Execution state | Issues, priorities, projects, ownership. |
| Notion | Human workspace | Visual browsing, lightweight pages, collaboration. |
| Slack / Granola | Capture | Fast intent, discussion, meetings, raw notes. |
| Memory + Skills | Personalization + reusable procedure | Preferences go in memory; workflows become skills. |
Meeting notes and transcript capture the messy conversation.
Human intent becomes a command or request to the agent.
Grounds itself in prior context before acting.
Decisions, tasks, and docs move to the right surface.
Agent reports what changed with links or diffs.
Markdown is inspectable. Diffs are reviewable. PRs create a human approval checkpoint.
zerogen-os: AGENTS.md, source-of-truth rules, simplified docs, PR history.
Inspects files, proposes structure, edits docs, opens or updates a PR.
Reviewed markdown becomes the operating manual future agents read.
Show ZeroGen Agent, ZeroGen OS, projects, issues, owners, priorities.
Show the human-facing ZeroGen pages: meetings, areas, lightweight overview.
The agent routes each question to the surface designed to answer it.
Read context, draft docs, summarize, propose changes, create artifacts, open PRs.
Merge PRs, change product canon, mutate priorities, delete durable context, make irreversible external changes.
Build environments where agents know where truth lives, what they can change, when to ask, and how to improve the next run.
Meeting → decisions → issues → docs.
Context librarian and drift detection.
Agent swarms for specs, QA, and review.