Agent Mondays: a help center that keeps itself current

Your help center is probably out of date right now. And the fix shouldn't require a single person's time.

At Perch, our help guides mattered. Prospects used them to evaluate the product. Customers referenced them when they got stuck. New hires leaned on them during onboarding. They were a real source of truth.

But the traffic was really low 😂 Most people, when they had a question, just called CS or pinged a salesperson directly. So the help center sat in this awkward middle ground: important enough that letting it go stale was a bad look, not valuable enough to justify dedicated time.

The burden fell on Product and CS to keep it current.

That was the real problem. Not the stale docs. CS was spending hours updating help articles instead of talking to customers. Every hour on documentation was an hour not spent on inbound questions, renewals, or onboarding calls.

I didn't solve this at Perch. But I carried the problem with me. On some side projects I've been building recently, I finally built the workflow I wish we'd had.

Here's how it works:

1. Every pull request triggers a read-only documentation review
2. The agent examines the code diff and identifies customer-facing changes
3. It checks those changes against existing help articles
4. Playwright signs into the product using a dedicated documentation account with stable sample data
5. It opens the affected screens, captures fresh screenshots, and checks whether existing visuals still match the product
6. The agent flags stale articles and screenshots, or proposes replacements directly in the pull request
7. Once the code reaches production, a write-enabled agent updates the relevant instructions and images
8. It publishes the changes and creates a short human-review task with the live pages, screenshots, and underlying commit
9. A daily reconciliation routine catches anything the event-driven checks missed
10. Heartbeats and health monitoring alert us if any part of the workflow stops running

Documentation maintenance no longer depends on someone remembering to do it. Product changes automatically become inputs to the documentation workflow.

The shift wasn't really about keeping docs current. It was about where CS spends their time. When the help center updates itself, your support team stops being editors and goes back to being what customers actually need: responsive humans on the other end of the line.

This is not a groundbreaking use case of AI, but you add up enough of these. wins and this can have a huge impact on your business or your employees' day to day.

This is Agent Mondays. Every week I am trying to share one concrete way I'm using AI to remove low-value work so your best people can focus on the work only they can do.

Is there a task on your team right now that a senior person is doing mostly out of habit, because no one ever stopped to ask if it still needed a human?