The work behindthe automation.
Stories from teams running Greyflows builds in production. Four delivered systems, each scoped, priced and shipped in under a week. Figures are client-tracked; quotes shared with permission.
Case studies, named with permission
Every build, with the numbers attached
How Verxi built Manual qualification 3h/day down to 12 minutes
“We were drowning in demo requests. Now the right leads hit Slack instantly with context.”
How Treda built First response 4h down to 11 minutes
“Treda finally has a defensible oversight trail for every ticket.”
How Obsyn built 2 silent failures caught in the first week
“We now see pipeline health before users do.”
How Notion Knights built Content creation 10h/week down to 45 minutes
“Our socials now run themselves. We just review and approve.”
Full detail
RAG pipeline observability
“We now see pipeline health before users do.”
Problem
Obsyn’s RAG pipeline had silent failures: embedding drift and latency spikes were not caught until users complained. There was no centralized telemetry or alerting.
What we built
A 12-node pipeline in two chains. The watch chain samples telemetry every 15 minutes, flags drift above 0.08, p95 above 1200ms or error rates above 1%, logs to the database, archives to Sheets and alerts Slack with a cooldown to stop storms. The report chain sends a daily 8am rollup of the last 24 hours of metrics to the team inbox.
Stack
Results
- 2 silent failures caught in the first week
- Mean time to detect down to under 15 minutes
- Daily rollup saved 4 hours/week of manual reporting
What clients say
Simple enough to see.
Powerful enough to ship.
Join the teams running automation they can actually explain.