Quality, at agent speed.

Yes to both. Starter includes a 14-day free trial: cancel before it ends and you won't be charged. After that, cancel anytime; your plan stays active until the end of the current billing period. No penalties, no lock-in.

AI test generation is built in. No setup needed. Point it at a Jira ticket or paste a description and SmartRuns returns structured test cases with preconditions, steps, and expected results in seconds. Enterprise customers can connect their own AI provider for full control over model choice and data routing.

Yes, reach out and our team will handle it for you. We'll scope it together based on your current setup.

Jira, GitHub, Slack, Notion, outbound webhooks, and a live BigQuery data connector for your own SQL and dashboards — plus a hosted MCP server at mcp.smartruns.io so any AI agent, running any model, can read and write your tests, Specs and runs directly.

SmartRuns runs on AWS with encryption in transit (TLS) and at rest, and every account's data is isolated from every other. Each change is captured in an audit trail, queryable in full through the BigQuery connector beyond the in-app retention window. If your compliance team needs documentation or logs, we'll provide what we have. Enterprise plans add a dedicated cluster and configurable sign-in methods.

Yes. Team and Enterprise plans include role management. Team lets you customise the default roles; Enterprise adds full control over sign-in methods and access configuration.

TestRail and Zephyr are per-seat test management built for manual QA — Zephyr Scale even bills for every Jira user, including people who never open it. SmartRuns is an agentic quality platform: the test repository, plans and runs you expect, plus Specs, a configurable agentic workflow, and MCP so your AI agents work in it directly. And it's a flat price with unlimited users — no per-seat math — self-serve in under 5 minutes.

You define your delivery process in SmartRuns — ordered stages, the agent role that owns each phase, your conventions, and where a human signs off. An AI coding agent (Claude Code, Cursor, or your own) connects over MCP and pulls that workflow live, then works through it, drafting Specs, tests and defects as it goes. Each draft waits in 'awaiting confirmation', and the human gates you placed hold until someone approves. Every step is recorded — which agent, which person, what changed. Update the workflow and every project picks it up on its next run.

AI-run work, like AI test generation or an agent-run implementation task, is never persisted automatically. Every AI task sits in an 'awaiting confirmation' state, and nothing it produced becomes real until that task is explicitly confirmed. Where a person has to be the one confirming — before a release, before a Spec closes — is a sign-off gate you configure in your workflow, so the human stays in the loop exactly where you decide they should be.

Yes. Every action in SmartRuns' audit log is attributed to one of three sources, never blended: a User acting directly in the app, an integration token (for example, an MCP call from Claude Code), or the system itself. So a change a person made in the app is always distinguishable from one that came through an integration — and it's the integration token that flags an agent's work, since that's the channel agents act through.

No. SmartRuns started as a complete test management platform (test repository, plans, runs, and AI-generated tests from Jira), and that still works perfectly on its own. Specs and the governed sign-off workflow are an extra layer for teams that want end-to-end traceability. Add it when you're ready, or never. It's not a requirement.