Fergus ships 6.3× more weekly merges with the same-sized team, cutting PR merge time from 18 hours to 1.6
Fergus engineering lead Ben Gracewood writes that the same-sized team as 18 months ago now merges 6.3× as many pull requests per week across about 150 GitHub repositories, including a core PHP monolith (deploys there up 250%) and surrounding services. Time to merge fell from 18 hours to 1.6 hours, which he attributes mainly to AI reviews and more ownership over smaller services. He also reports engineering throughput up 500%, app performance tripled, and change failure rate halved, while modernising a decades-old legacy stack (ClickHouse reporting APIs, OpenSearch, routing and auth refactors, a LangGraph-on-Aegra AI platform, and PHP modernisation). The post does not name which coding agents or review tools they use.
The numbers (as reported)
| Weekly code merges | 6.3×, same sized Fergus Engineering team |
|---|---|
| Time to merge PRs | 18 hours → 1.6 hours |
| Engineering throughput | Up 500% |
| Monolith deploys | Up 250% |
| PRs per author | 4.7× |
| App performance | Tripled |
| Change failure rate | Halved |
| Team size | Same as 18 months ago |
| Repositories | About 150 GitHub repos, including a core monolith |
Facts
- Company / subject
- Fergus
- Tools
- Unstated
- Codebase
- brownfield
- Work type
- Company rollout
- Stack
- PHP, GitHub, ClickHouse, OpenSearch
- Published
- 2026-11-09
- Last checked
- 2026-11-10
What the source claims Practitioner (own blog/paper)
Claimed by the builder, writing about their own work. Not checked independently.
- 6.3× overall code merges per week with the same sized team 6.3× weekly merges
“6.3x overall code merges per week, with the same sized Fergus Engineering team.”
- Time to merge PRs fell from 18 hours to 1.6 hours 18 hours → 1.6 hours
“Our time to merge PRs has plummeted from 18 hours to 1.6 hours, mainly driven by AI reviews and more ownership over smaller services.”
- Engineering throughput up 500% Up 500%
“Engineering throughput up 500%, merges are 10 times faster, app performance tripled, and change failure rate halved.”
- Change failure rate halved, despite a handful of outages Change failure rate halved
“We have caused a handful of outages, but our change failure rate has halved.”
- 4.7× pull requests per author 4.7× PRs per author
“6.3x the PR rate overall, 4.7x PRs per author.”
Caveats
Self-reported by Fergus's engineering lead. The post does not name which coding agents or AI review tools they used. Throughput (up 500%) and weekly merges (6.3×) are related but separately stated; raw PR counts are not given. 6.3× overall vs 4.7× per author is explained as more merged PRs coming from outside the engineering team. They report a handful of outages even as change failure rate halved. Comparison window for the 6.3× figure is not dated beyond the same team size as 18 months ago.
More stories by tool: Unstated
Related topics: Company rollouts and costBrownfield codebases