Hermes
Available nowFast, stateless, built to be called
A lean harness for agents that run on demand. Low cold-start, no session state to manage, and a clean surface for wiring into existing systems.
Best for
Workloads that fire an agent per request (pipelines, webhooks, internal tools) where predictable latency and clean isolation matter more than continuity.
Hermes is the opposite shape from a persistent employee. Nothing carries over between runs, which makes it straightforward to reason about and easy to scale sideways.
What it still needs is somebody watching capacity, keeping cold-starts inside the envelope, retaining traces long enough to debug a bad run, and shipping upstream harness updates without breaking the contract your callers depend on.
Managed plans
What it costs to have us run it
Starter
$49per month
- 25,000 agent runs per month
- 7-day trace retention
- Automatic harness upgrades
- Email support
Scale
$290per month
- 250,000 agent runs per month
- 90-day trace retention
- Dedicated capacity with per-tenant limits
- Staged upgrades with rollback
- Shared Slack channel
Need something outside these shapes? Talk to an operator
Other harnesses we run
Same operating layer, different shape of work. Pick the one that fits.