The hardest part of support isn't the fix. It's finding the problem in a pile of logs.

The inbound data dump

We largely use our support portal, but occasionally customers email us directly — big zip files with densely packed logs, links to GitHub issues, and whatever other context they think is relevant. There's a lot of work that goes into organizing the information users give us so we can actually begin doing the job of trying to understand what went wrong.

Being able to use TextForge means Claude Code can grab the email, download all the attachments, unzip them, grep through the contents, and extract what I think of as keyframes — the key pieces of detail I need to take a look at in order to begin my analysis.

It's not doing my thinking for me, but it is sifting through what's usually a fairly unorganized dump of data and helping me set some really good directions right out of the gate.

The outbound summary

The other angle is what happens once I've come up with a fix. If there's a long GitHub issue thread or a multi-person email chain, Claude helps me summarize the resolution and draft a concise email back to all the principals on the original thread. They get a clear, professional reply without me spending 30 minutes writing up what I just spent an hour diagnosing.

How it works