A real-world MCP debugging win with Better Stack
Company: Upheal
Uphealis a smart clinical platform and AI-native operating system designed for mental health professionals. At its core, Upheal generates automated AI progress notes and clinical insights for every session, whether used as a specialized documentation assistant alongside an existing EHR or as a complete practice solution.
For private pay practices, UphealOS serves as a comprehensive EHR replacement, integrating scheduling, client payments, and shareable forms into a single workflow. By automating the administrative and documentation burdens of therapy, Upheal allows clinicians to stay fully present during sessions and reclaim precious time for their clients.
Industry: Mental health care
Use cases: MCP server, Telemetry, Uptime monitoring, Status page
Switched from: AWS CloudWatch
From Uptime to centralized telemetry
Upheal originally started with Better Stack Uptime and later migrated their logs fully from AWS CloudWatch to Better Stack Telemetry. Onboarding was straightforward, and the team quickly standardized on Better Stack across their entire stack. “Your UI is already very useful when troubleshooting issues,” says Martin, co-founder & CTO at Upheal. Dashboards, alerting, and Slack integrations made it easy to stay on top of production health.
When logs were not enough
As Upheal scaled, the team began encountering intermittent production issues that were difficult to reproduce. Traditional log searching helped, but spotting subtle patterns across services and code still took significant 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,” Martin notes.
MCP and AI-powered observability
Curious about new AI tooling, Upheal tried Better Stack MCP after spotting it in a recent product update. The idea was simple but powerful: give Claude access to production logs and see what happens. MCP allowed them to combine their logs with their codebase and run intelligent analysis directly on real production data.
"We were wondering what happens if we give Claude access to logs."
— Upheal Team
One long-running issue involved intermittent server timeouts that had resisted multiple investigations. Using Claude Code with Better Stack MCP, the team analyzed logs and cross-checked them against their code. After a few iterations, Claude uncovered a pattern the team had missed and traced it back to a niche configuration issue in a library.
What previously would have taken hours of manual digging was resolved in minutes.
From hours to minutes: A new standard for production troubleshooting
- Faster root cause analysis for intermittent incidents
- Fewer escalations and less time spent searching logs
- One reliable platform for uptime, log management, dashboards, and incident response
"What would have taken hours before now takes minutes. It’s a game changer for us."
— Upheal Team
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.