Why Preset Needed Help
Preset runs the managed service for Apache Superset. The open-source BI project has exploded to more than 66K GitHub stars and a community of thousands of users and contributors. Keeping up with that traffic was stretching Preset’s small maintainer team thin.
The Fix: Dosu Inside GitHub
Instead of adding head-count, Preset wired Dosu into both GitHub Issues and Discussions. The bot works around the clock, reads Superset’s code and historical threads, and answers in multiple languages. Maintainers stay in the loop, approving or editing replies when needed.
Results After Go-Live
- Lower triage load. In the first 2600 new issues after install, Dosu handled 66% of the sorting and first responses.
- Faster answers. Median time to first reply dropped from about two days to under five minutes, a pace that’s hard to match by hand.
- Happier contributors. Community members now get actionable replies, or linked duplicates right away, so real problems surface faster.
What This Looks Like From the Dosu Side
Open source gives Dosu a full view of the code, past discussions, and merged PRs, so it can spot patterns a human might miss. That context lets the bot:
- Flag likely duplicates before they bloat the backlog.
- Suggest config tweaks for edge-case deployments.
- Point newcomers to contributing docs the moment they hit “New Issue.”
For Preset, that translates to fewer repetitive questions and more bandwidth to ship features like dashboard improvements and Superset 5 compatibility.
Where We’re Going Next
Preset and Dosu are already testing AI-generated release notes and automated doc updates pulled from merged PRs. Longer term, we’re exploring AI-assisted bug fixes that open a draft PR with a proposed patch—maintainers stay in charge, but routine work gets a head start.
Read the Deep-Dive
Want the full story, numbers, and screenshots? Check out Preset’s detailed write-up on their blog. Then, if you run an open-source project with a growing backlog, see how Dosu’s open-source program can lighten the load.