Key Takeaways

  • Ryan Carson runs Untangle as a solo founder, managing customer success, sales, and engineering by deploying 10 to 15 autonomous Devin agents in the cloud.
  • Carson ships roughly 40 PRs every day, creating a massive verification bottleneck when tracking which bugs are active, in review, or already fixed.
  • Untangle serves family law firms handling sensitive caseloads, making manual triage across every client account impossible for one person.
  • Carson created an autonomous diagnostic agent routine to scan client telemetry, isolate top errors, and check GitHub pull request statuses.
  • This operational system is codified in Carson's Watchdog Account Triage Playbook.

The Carson's Watchdog Account Triage Playbook

  • Step 1: Account-by-Account Scan: Execute a scheduled or on-demand scan across every customer firm account to assess recent user activity and system health since the previous run.
  • Step 2: Error and Telemetry Ingestion: Pull active error logs, UX breakages, and Sentry reports associated with each specific customer account.
  • Step 3: Synthesis and Prioritization: Filter and transpose raw telemetry data into a prioritized summary listing the top three critical issues affecting that account.
  • Step 4: PR Status Cross-Referencing: Cross-reference identified issues against the repository to determine if a fix has merged, is in an active PR, or remains unaddressed.
  • Step 5: Interactive Thread Resolution: Review the synthesized summary in the agent thread and spin off targeted bug-fix tickets directly from the findings.

When This Works (and When It Doesn't)

This workflow is essential for solo founders or small engineering teams scaling B2B accounts who need automated customer success and proactive issue resolution without human account managers. Carson built this because rapid AI code generation creates visibility blind spots. As Carson explains, “We're starting to pick up a lot of family law firms now and they have a lot of cases. And I'm literally doing customer success and engineering and sales. So it's like, whoa, how do you keep track of actually what's going on in these cases?”

When stress spikes and operational clarity vanishes, automated triage replaces panic. Carson notes, “So what you want is that moment where you feel like, 'I need to know what's going on in the business right now. I feel stressed. I don't know what's happening.' This is when I run Watchdog.” The agent gathers account logs and presents an unvarnished audit. As Carson puts it, “You don't want just data. You want it to be transposed and filtered and sorted for you. So then it comes up with the top three things that are going wrong.”

High code velocity complicates tracking. Carson points out, “Did it get fixed or is it currently being fixed or is there an open PR that never got merged? It gives me all that and then I talk to that thread about okay, let's dig in a little bit.”

Where this breaks down is on subtle user experience issues that never trigger Sentry events. If a workflow confuses a lawyer but does not throw an uncaught exception, Watchdog will report a clean bill of health. High-touch customer conversations must still validate whether the product actually solves the client's problem.

What to Do With This

If you are a solo builder managing multiple client accounts, build a scheduled script that runs against your Sentry and GitHub APIs every morning at 8:00 AM. Configure the script to iterate through your organization IDs, pull all errors logged in the past 24 hours, and rank the top three by user impact per account. Next, match the error stack traces against recent commit messages and open pull request titles. Have the script format this synthesis into a single Slack digest or agent thread before you start your customer calls.