Complete your Databricks User Groups profile!

Fill out a few details about yourself so the community can get to know you.
Genie Community

Following up on the rollout post: what actually built trust in Genie wasn't accuracy. It was showing people the answers were being watched. 🤝

Summary: Valeria Kozel shares their team's approach to building trust in Genie Rooms by emphasizing accountability and transparency rather than just accuracy. They describe using a feedback loop with thumbs up/down mechanisms, ensuring any feedback is addressed by the product team, highlighting the importance of seeing real changes. Matthias Mueller raises a concern about potential bottlenecks similar to those in central data warehouse teams. Valeria Kozel clarifies that accountability doesn’t obstruct access, but ensures improvements are tracked and implemented effectively. Matthias Mueller appreciates the clarification and inquires how to encourage user feedback effectively.
AI Summary

A Finance team won't trust a Genie Room just because it's right. They trust it when they see someone is accountable for the answers.

What worked for us:

  1. a thumbs up/down on every chat, and total transparency about who reads it. We had a dedicated Finance team owning the rollout — they knew the real questions and investigated every miss.

  2. and a thumbs-down doesn't just sit in a dashboard; it goes to the product team, and Databricks genuinely comes back to you on it.

That's the shift: trust isn't won with accuracy stats — it's won when people see a loop. My feedback → a real person → a real change.

How do you close the feedback loop on your Genie rooms — and who owns the answers? 👇

#Databricks #Genie #ChangeManagement

4 comments
Matthias Mueller
1 week, 5 days ago

But with this approach you build the same bottleneck as in existing centralist data warehouse teams.

No person available -> no data to trust and publish -> still slow

Valeria Kozel
1 week, 5 days ago

Spot on to flag it, Matthias — worth naming explicitly — I'd push back gently on one thing: accountability doesn't have to mean a person in the path.

The bottleneck you describe comes from humans gating access. What I'm describing is humans owning fixes. Access stays fully self-serve — anyone can ask the room anything, anytime.

What's accountable is that when an answer's wrong, someone makes sure it's fixed in or added to the room (instructions, examples, metrics), not patched by hand each time. Conflate the two and you've rebuilt the central team; eep access open, make improvement accountable, and you avoid it (i.e., rebuilding the central team).

Does that hold up against what you've seen?

Matthias Mueller
1 week, 5 days ago

Ah, Valeria, now I get it and it matches with what we are currently establish within our company.

How have you achieved it that user give feedback (thumps up/down or request review) so the responsible team can have a look at the response and fix it, if needed?

Valeria Kozel
1 week, 4 days ago

The trick for us was around two things: have a clear community who owns the feedback and ensuring the fix is easy to do:

We created a small group of sponsors and champions — users who already love Genie and want it to win. They triage each question into the right layer (instruction, example, or metric), prep the fix, and push it to prod at the next release.

And because most fixes are simple — a tweaked instruction, an added example — almost any user can do them, so it never bottlenecks on a central team.

People give feedback once they see it actually change the room.

Valeria Kozel
2 weeks ago