Life360, a family connection and safety company with nearly 88 million monthly active users, wanted to scale AI responsibly across IT and engineering without creating new complexity or sprawl. Early attempts at single-purpose bots and broad, do-everything agents ran into issues: individual bots left employees unsure which one to ping, while a single overstuffed agent became slow and shallow. In both cases, users fell back to pinging IT directly.

IT ticket handling was another challenge. A Jira ticket could sit for six hours before anyone on IT picked it up, creating a window where an agent could add value by preparing context and starting troubleshooting. Slack channels also ballooned to more than 2,500, with no lifecycle controls in place. “I wanted to make sure we didn’t build tech debt from day one. People forget which bot to ping and go back to pinging IT. We needed one place to ask, and let it figure out which sub-agent to use,” said Matt Currie, Automation Engineer at Life360.

Life360 built Mimir, named after the Norse god of knowledge and wisdom, as an orchestrator agent on Tray Merlin Agent Builder. Employees @Mimir in Slack with their requests, and Mimir routes them to specialized sub-agents for ITSM, Jira, Confluence, Slack management, coding, and mobile crash detection.

The team is also developing a Crash Agent designed to identify an app crash, locate the code path, propose a patch, and open a pull request for engineers to review. “You just @Mimir and tell it what you need. It figures out which sub-agent to engage,” said Matt.

Governance and reliability are built into the design. New features launch first in a sandbox agent called @DRADIS, visible only to IT and security for testing, then move into Mimir once validated. If one LLM provider is down, Mimir fails over to another so responses continue uninterrupted. Slack hygiene is automated through Tray workflows that audit thousands of channels, notify owners, and archive inactive ones, so the Slack agent can safely create new channels on user request. The next step is Jira and email ingestion so tickets can be pre-worked before IT’s SLA.

Life360’s agent-of-agents model is already returning time to IT and laying a scalable foundation for AI across the business. Employees now have a single entry point in Slack, eliminating confusion over which bot to contact. Jira tickets created through Mimir contain full Slack-thread context, improving handoffs and reducing back-and-forth. Slack lifecycle automation, powered by Tray workflows, regularly audits thousands of channels, notifies owners about inactivity, and archives unused spaces, giving Life360 confidence to allow self-service channel creation.

• One front door in Slack reduces agent sprawl and “who do I ping?” friction
• Ticket pre-work adds troubleshooting and context before IT’s ~6-hour SLA window
• Slack lifecycle automation across 2,500+ channels makes it safe to allow user-requested creation
• Early Crash agent testing demonstrates the ability to generate draft pull requests that can save engineer hours
• Model failover ensures reliability during provider downtime

“The moment it clicked was when I saw an agent run end-to-end and open a pull request in the repo. It showed me we could actually close the loop on issues, and that’s real value,” said Matt. “We’ve only just started, but we’re going full steam.”