How Upheal found a timeout bug in minutes by giving Claude access to its logs
Upheal moved its logs off AWS CloudWatch and standardized on Better Stack, then cracked a run of intermittent server timeouts in minutes by giving Claude Code access to production logs through the Better Stack MCP server.
- Industry
- Mental health care
- Products used
- Telemetry, uptime, status pages, MCP server
- Switched from
- AWS CloudWatch
From uptime checks to full telemetry
Upheal started with uptime monitoring and later moved its logs off AWS CloudWatch entirely. Onboarding was straightforward, and the team standardized across the stack, with dashboards, alerting and Slack integrations keeping them on top of production health.
The bug that survived every investigation
As Upheal scaled, intermittent production issues started appearing that were hard to reproduce. Log search helped, but spotting a subtle pattern across services and code still took real time and effort.
“We were hitting a wall with some intermittent server timeouts and spent a lot of time trying to track them down without much luck.”
What happens when you give Claude the logs
The team spotted the Better Stack MCP server in a product update and tried it, mostly out of curiosity. The setup was simple: Claude Code connected to production logs through the MCP server, with the codebase alongside it.
After a few iterations, the analysis surfaced a pattern the team had missed and traced it back to a niche configuration issue in a library. What had resisted hours of manual digging was resolved in minutes.
What changed
Root cause analysis on intermittent incidents is faster, fewer issues escalate, and less time goes into searching logs by hand. Uptime, log management, dashboards and incident response all sit in one platform.
“What would have taken hours before now takes minutes. It's a game changer for us.”
Uptime, logs, metrics, traces and incidents in one platform, with an MCP server for your AI tools.