The whole company asks one coworker.
Spur runs customer conversations for online stores. Its support and success team, its founders and its engineers all mention Doozy in Slack, and the questions that used to wait on an engineer come back answered in the thread.
Spur has handed work to Doozy since August 2026: first the build that became the product, then the product itself from 20 October. Figures read on 26 October 2026.
- Company
- Spur, spurnow.com
- Who asks
- Support, success, founders, engineering
- Works in
- Slack, GitHub, Stripe, Linear, the production database
- Since
- August 2026
Since August, in numbers
Read from Spur's own records on 26 October 2026. Where a count is a floor, it says so; nothing is rounded up.
- 2,000+
- jobs done
- 1,400
- threads answered
- 478
- scheduled runs
- 1.6 min
- median ask to answer
a floor: every job that changed Doozy's memory, plus every scheduled run
across 42 Slack channels
21 schedules active today; 440 of 478 runs succeeded, 29 skipped, 9 failed
on the product, 20 October 2026 to 26 October 2026
On the product itself, in its first week (20 October 2026 to 26 October 2026): 50 jobs, 49 done and one failed and said so; 7 people asked; 15 runs on a schedule, median 3.1 minutes.
What Spur hands over
The same coworker, three kinds of work. Each is asked in a thread, after @Doozy, and answered there with what it did and where each fact came from.
Support and success
- The root cause of a customer complaint, found in the product and the database before anyone from engineering is pulled in.
- Billing forensics: what a customer was charged, when and why, checked in Stripe against the account.
- CSV exports of conversations, accounts or orders, in the shape the person asked for.
- A drafted reply to the customer, from the conversation and what the account shows.
- A Linear ticket filed with the examples attached, when the cause is a bug.
Founders and engineering
- Data pulls and pricing analysis from the production database, with the query shown.
- Pull requests opened for small fixes, with tests run and the diff in the thread.
- Skills written for the team: ways of working Doozy follows on the next ask.
On a schedule
- The daily support-queue report, posted to the support channel every morning.
- A daily review of the day's customer calls.
- The weekly product update, assembled from what shipped.
In Spur's words
Three teams, three asks. Approved for publication by Spur.
"A billing question used to mean a ticket to engineering and a wait. Now I ask in the thread and get the Stripe history, the root cause and a drafted reply back before the customer has replied. The morning queue report runs itself, and it's much easier to see what matters first."
"Before a call I ask what an account has actually been doing, usage, billing, open tickets, and it comes back in the thread with where each fact came from. Then I ask it to log the call and set the next step. The CRM finally stays honest without me living in it."
"It's one hire the whole company can ask. Support gets root causes without pulling an engineer, we get the week's numbers from our own database with the query shown, and the small fixes come back as pull requests. We set it up ourselves and it has stayed inside the lines we gave it."
How it is set up
Spur's database is the company's source of truth, so the setup is the one the security page describes, with nothing special.
- The production database is connected read-only. Doozy can query it and can't change it.
- Every task runs in its own isolated workspace. Credentials stay outside it.
- On GitHub it opens pull requests on a branch. It can't merge or push to main.
- Tools that delete or overwrite start off. Every action is on the audit log.
Your next hire is one mention away.
Add Doozy to Slack and hand it the thing that's been waiting on someone all week. It comes back done, and it starts learning how your team works from the first ask.
- 14 days free, no card
- One price for the whole team
- Nothing billed until you pick a plan