Hermiq
Schedule autonomous AI agents in . Give an agent a prompt, tools and skills, then run it daily, hourly, or on a flow trigger — delivered straight to Nextcloud Talk.
Autonomous agents, on your own Nextcloud.
Hermiq turns a prompt, a set of tools and a schedule into a working agent. No external agent hosting, no separate vector database — agents, their memory and their skills all live as OpenRegister objects inside your own Nextcloud instance.
Agents with tools and skills.
Define an agent with a prompt, a model, MCP tools, and installable skills from the agentskills.io catalog. Manage them from the Agents catalog and detail page.
Scheduled runs, delivered to Talk.
Attach a schedule to an agent — daily, hourly, or flow-triggered — and its output lands in a Nextcloud Talk conversation (or a notification when Talk isn't installed).
Chat, sessions and memory.
Talk to any agent over a live streaming chat thread. Sessions keep the full conversation history; consolidating memory keeps long-running agents within their context budget.
Human approval gate.
Route sensitive agent actions to a reviewer inbox before they run. Approve or deny per action, with an org-level kill switch.
Governed by OpenRegister.
Every agent, run, memory and skill is an OpenRegister object — audited, RBAC'd and multi-tenant, with a run-analytics view and a per-org quota and audit export for tenant admins.
Partners shipping hermiq
Implementation, hosting, and integration partners that deliver hermiq to their customers. Pick the partner you already work with, or one whose stack matches yours.