agent infrastructure
Run AI agents
in production.
Demos are easy; Tuesday at 3 a.m. is hard. Trellis is the runtime between your agents and your systems: durable orchestration, guardrails the model can't talk its way past, eval gates in CI, and a priced trace for every run.
- First run traced in ~10 min
- Free dev tier · 1,000 runs/mo
- $0 margin on tokens
Engineering teams running agents on Trellis
Sample customer names. Every company here is fictional.
- Ledgerwick
- Fenwright Genomics
- Bracken Freight
- Quillbase
- Mosstone Labs
- Vantry
the loop
Orchestrate. Guard. Evaluate. Observe.
Four modules, one runtime, one trace model. Each is useful alone; together they close the loop that keeps an agent fleet improving instead of drifting.
01 / 04
Orchestration
Durable runs that survive restarts, retries, and people
Explore orchestration →
02 / 04
Guardrails
Policy checks at run and call boundaries, enforced by the runtime
Explore guardrails →
03 / 04
Evals
Regression gates for behavior, wired into CI
Explore evals →
04 / 04
Observability
Every run a trace; every span priced
Explore observability →
observability
Every run is a trace. Every span has a price.
Model calls, tool calls, guardrail verdicts, and approvals land as spans in one tree, with token counts, latency, and dollars on every node. When a run goes sideways you read what happened; you don't rebuild it from logs.
The same trace data becomes your eval sets, your cost rollups, and your invoice line items. One source of truth, three jobs.
See the trace modelPlatform figures
sample figures · fictional company
- 11 ms
- median orchestration overhead per agent step
- 99.96%
- runs API uptime
- 38M
- agent runs traced on our peak day
- $0.00
- our margin on model tokens
p50, all regions, trailing 90 days
trailing 12 months, public status history
sum across customer workspaces, June 2026
pass-through at provider list price, invoiced line by line
developer experience
Agents as code, budgets as law.
Declare tools with argument schemas, attach policies, and set hard budgets on tokens, spend, and wall-clock time. The runtime enforces all of it. The step that would blow the budget never executes, and the failure path is a branch you wrote, not an exception you didn't.
TypeScript and Python SDKs, an HTTP API underneath, OpenTelemetry out the back.
a note before you buy
Most workflows shouldn't be agents. We wrote that down.
If your workflow is enumerable, a pipeline beats an agent on cost, latency, and debuggability, and we'd rather you find that out before an invoice. Our CTO's essay has the decision framework and the arithmetic.
first questions
What engineers ask on the first call.
Short answers here. The FAQ covers pricing, data handling, and running in production in more depth.
Do I need an agent, or would a simple pipeline do?
If someone could write the steps down in advance, build a pipeline. It will be cheaper, faster, and easier to debug. Agents earn their cost only when the next step depends on what the last one found. Our essay on when not to use an agent walks through the test.
How does pricing work?
Two meters and a pass-through. You pay a small orchestration fee per run and $0.08 per 1,000 guardrail checks. Deeper trace retention is a flat add-on per run. Model tokens pass through at provider list price with no markup.
Will you go through our security review?
Yes. We fill in your questionnaire, share a data-flow diagram, and walk your team through access, retention, and key controls on a call. This sample site does not claim any audit or certification.
$ trellis estimate --honest
Price your stack before you talk to us.
The configurator does the arithmetic in the open: per-run cost at 10k, 100k, and 1M runs, assumptions stated. Bring the estimate to a demo, or don't book one at all.