Current

Planned

considered

Roadmap

Fine-grained control over what each job and agent can access. Scoped permissions per job, per tool, per data source.

Considered

Granular permissions and access control

Fabian Both

Go beyond schedules. Trigger jobs from webhooks, external events, or other systems.

Webhooks and event triggers

Let other agent frameworks like OpenClaw create and manage Golemry jobs via MCP or CLI. Golemry becomes the reliability backend for any agent stack.

MCP/CLI for external agent frameworks

React to overseer decisions: approve requests, comment on escalations, or give feedback on past calls. The overseer adapts its gating based on your input, improving its judgment over time.

Learn from your feedback

One dashboard for all job results, overseer notifications, and execution metrics. See what ran, what passed, and what needs your attention.

Agent inbox and monitoring

A dedicated AI validates each job's success, guarding actions, and potentially escalating problems to the user. When confidence is low, it escalates instead of shipping.

AI overseer for every job

Describe what you want automated in a chat, and Golemry creates a cloud job that runs on your schedule. Set up context, instructions, and triggers through natural conversation instead of configuration.

Scheduled AI jobs from conversation

Every job runs in an isolated cloud environment. No local setup, no infrastructure to manage.