Recursion

What it does

Recursion Managed Agents is a hosted platform for running AI agents inside business applications such as Salesforce, Zendesk, NetSuite, Slack, Jira and Datadog. Agents start from a schedule, a message, an alert, a merged pull request or an API call, then gather information from connected tools and carry out multi-step work. Depending on the workflow, outputs can include pull requests with tests, incident root-cause drafts, postmortem drafts, dataset audits and renewal briefs.

People use it to set up recurring or event-driven jobs, such as an on-call incident coordinator that checks logs, deploys and feature flags after an alert, or a dependency upgrader that opens a pull request per package release. A coordinator splits work across specialist agents running in parallel, and a separate grader checks each session against a rubric before results are accepted. Finished sessions write short reports that the agent reads as memory on later runs.

The product suits engineering, security, research, data and operations teams with repetitive, queue-driven work. Sandboxes, credentials, subagents, memory and grading are handled by the platform. Access is through the vendor's 'Contact us' or 'Get started' links. Agent actions are limited to the tools an organization enables for each agent.

What it can do

• Runs agents on schedules, events, alerts and API calls
• Coordinates parallel specialist agents under a coordinator
• Grades each session against a customer-defined rubric before accepting results
• Stores session reports as memory that later runs read before acting
• Browses web portals with computer use, saving screenshots of each step
• Provides a fleet dashboard showing session status and grades

Works with

• Salesforce
• Zendesk
• NetSuite
• Slack
• Jira
• PagerDuty
• Datadog
• GitHub
• LaunchDarkly
• Stripe

Example workflows

Incident triage on a failing deploy

Input: a PagerDuty alert and a deploy at 14:05. Steps: a coordinator reads the alert, error graphs and deploy, plans five checks, and specialists review logs, the deploy diff and feature flags in parallel. Output: a mitigation status and a draft fix pending on-call approval, per the page's example.

Dependency upgrade per package release

Input: a new package version released in GitHub. Steps: an agent upgrades the dependency, fixes what breaks and reruns tests. Output: one ready-to-merge pull request with tests passing, per the page's workflow description.

Before you choose

• Agent actions are limited to the tools you allow; set those permissions before enabling agents.
• Pricing and trial terms are not published on the page; confirm them directly with the vendor.

Listed under

• Coding
• Code review and testing
• DevOps and observability
• AI agents
• Code review
• DevOps
• Engineering automation
• Enterprise
• Incident response

Get started

The page offers 'Contact us' and 'Get started'; the 'Get started' link points to a Recursion sign-up page.

1. Connect the business apps the agent should use, such as GitHub, Jira, Slack or PagerDuty.
2. Define what starts each agent: a schedule, an alert, a merged pull request or an API call.
3. Write a grading rubric so a separate grader can accept or reject each session's output.

The page does not state plans, usage rates, seat minimums or trial terms.

We couldn’t find published prices for Recursion. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Recursion Managed Agents | Labelbox ↗labelbox.com

Company website

labelbox.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.