Define active_customermerged
ManagedWe run your Context Factory
A SignalPilot forward deployed engineer keeps your dbt models, semantic models, and skills correct every week. They work your review queue, settle open questions with their owners, merge corrections, and run the evals. And they stand behind 90% precision on your graded questions.
- thumbs-down
Q2 net revenue thread · number disputed
- new question
How many active customers in Q2?
- eval failed
refund_rate_by_segment · run #133
- drift
dim_customers changed · 3 marts downstream
review queue · votes in · Mon 09:00
Refund rate by segmentsent back
Net revenue Q2 threadmerged
Every thread carries the query, the join grain, and the rows behind its number.
- ✓definition + regression testmerged to your repo
- ↺back to the queueneeds wholesale exception
- ✓context rulemerged to your repo
Week closed: 2 definitions, 1 regression test, precision 92.4% trailing 30 days.
What is included
The loop, run for you.
- Weekly review of coverage, failures, and drift.
- Exception queue worked: every pull request's recommended evals written and run.
- Every suggestion and thumbs-down reviewed within one business day, credited if it was our fault.
- Integration weeks on annual Scale and Enterprise plans: your BI tools, dashboards, and agent traces ingested, the dbt project fixed, evals written, nightly run wired.
The guarantee
Precision, measured monthly.
On the graded questions we froze together, trailing 30 days.
90%precision floor on your frozen suite
25–50%credit on that month's outcome fees between 80% and 90%
Refundedbelow 80%: the month's outcome fees returned. Model tokens are passed through at cost and are not refunded
WHO IT IS FOR
Teams whose coverage is below the floor, or teams with an experienced data lead but no capacity to take on another internal project.
WITH PARTNERS
Certified consultancy partners deliver the same service on the same tooling. See the partner program →