Starter
$29/ mo
For one team finding out what to automate first.
- 1 workspace
- 3 concurrent agents
- Full run history
- Community support
Relay plans a task, calls the right tool, and checks its own work — with a pause button on anything that should wait for you.
Plan
Lay the quarter out before anyone opens a doc.
Build
Ship changes in small, reviewable steps.
Launch
Tell everyone what changed, in plain language.
Connect your tools once. Relay's agents handle tickets, deploys and reporting on their own, and hand back exactly the decisions that need a person.
Why teams switch
Inputs, outputs and the exact prompt that triggered them, kept for every run an agent makes.
Set a limit once; a run stops itself before it crosses it.
Mark any action as "ask first" and it waits for a person, every time.
One agent can read a ticket, check a log and open a pull request in the same run.
Step back through exactly what an agent saw and did, in the order it happened.
Watch the running total the moment a call completes, not after the invoice.
Inside a run
Point it at the repo, the tracker, the docs. No migration, no new data lake.
Decide what runs on its own and what waits for a yes. Change it any time.
Plans, tool calls and cost show up in one timeline, live.
It hands back exactly the decisions that need a person, nothing more.
Sources
The same pipeline runs quietly behind every task Relay takes on.
Plan
Breaks the request into the smallest steps it can
Gather
Pulls in the tickets, docs and logs it needs
Act
Calls the one tool that moves the task forward
Check
Compares the result against the policy you set
Report
Hands back a summary, or waits for your go-ahead
Pricing
$29/ mo
For one team finding out what to automate first.
$79/ mo
For a team running agents against real tickets and deploys.
$199/ mo
For an org running agents across every team.
FAQ
Filter by what you're trying to figure out — billing, security, or how Relay fits your stack — then expand a question for the full answer.
Integrations
Connect your stack once. Each integration stays in sync on its own, so the run log is the only thing your team has to watch.