How Meridian Tech captured 8 years of knowledge in 72 hours
When their senior platform engineer gave notice, they had 2 weeks to capture critical deployment knowledge or risk months of production incidents.
The crisis
On a Tuesday morning, Marcus Chen — Meridian Tech's senior platform engineer — gave his two weeks' notice. He'd been with the company since the beginning. Eight years of building, debugging, and scaling their infrastructure. He knew every deployment script, every database quirk, every third-party integration.
The CTO immediately felt the weight of it. Marcus was the only person who could deploy to production without asking for help. Their deployment docs? A 6-month-old Confluence page with half the commands outdated and zero context about why things were done a certain way.
They had two junior platform engineers — both talented, but they'd never done a full deployment alone. They'd always shadowed Marcus. And now they had exactly 10 business days before he was gone.
- •Production deployment process (20+ steps, multiple failure modes)
- •Database migration procedures (including rollback scenarios)
- •Infrastructure debugging knowledge (why the cache behaves weird on Tuesdays, etc.)
- •Third-party API quirks and workarounds
- •Security protocols and access patterns
How they used Understudy
The CTO sat Marcus down with Understudy and ran five focused interviews over two days. Each session was 20-30 minutes, covering one critical area: production deployment, database migrations, monitoring and alerts, emergency rollback procedures, and infrastructure quirks.
Marcus talked through each process like he was training someone in person. Understudy asked clarifying questions: "What if the health check fails?" "How do you know when to roll back?" "What's the command for that?"
Understudy generated five playbooks. Marcus reviewed each one, adding edge cases and correcting a few commands. The junior engineers read through and flagged areas that needed more detail. Marcus jumped back into Understudy, answered their questions, and the playbooks updated automatically.
One of the junior engineers — Sarah — followed the production deployment playbook start to finish. Marcus watched but didn't intervene. She hit one spot where the playbook said "wait for cache warmup" but didn't explain how to check if it was done. Marcus added that detail. Sarah completed the deployment successfully.
With Marcus still around but deliberately hands-off, the other junior engineer — James — did a full production deployment using only the playbook. No questions. No errors. 47 minutes start to finish. Marcus was stunned.
The results
Marcus left on good terms. His last day was a normal day. No emergency handoff meetings, no frantic screen-sharing sessions, no "just call me if you need anything" promises that everyone knows create guilt and dependency.
The week after Marcus left, the team deployed three times. No incidents. No panicked Slack messages. The playbooks had everything they needed.
Two months later, Meridian hired a new senior engineer. Instead of spending weeks learning the deployment process, she read through the playbooks and was deploying independently within 3 days.
Time savings
Business impact
"I thought we were screwed when Marcus gave notice. We had two weeks to capture eight years of knowledge. Understudy made it possible. The playbooks it generated weren't just documentation — they were actually usable. Our junior engineers were deploying to production confidently within days. That would've taken months of shadowing and mistakes without it."
What Meridian learned
Don't wait until someone gives notice
After seeing how well it worked, Meridian immediately started capturing knowledge from other senior team members. "We're not waiting for the next crisis," Sarah said. "We're building playbooks proactively now."
AI interviews capture things written docs miss
Marcus said things in the interviews he'd never have written down. The edge cases, the "oh yeah, one time this happened..." stories, the tribal knowledge. Understudy's questions surfaced all of it.
Playbooks stay current when they're easy to update
When deployment processes change now, someone jumps into Understudy, does a 5-minute interview about what changed, and the playbook updates. Keeping docs current went from "impossible" to "actually happens."
Capture your team's knowledge before it walks out the door
Don't wait for a crisis. Build playbooks today.
Try Understudy free