For Data LeadsBe the agent orchestrator, not the bottleneck
Claude Code makes one analyst faster. Your job is getting the whole company to the same number. SignalPilot drafts the context as code and keeps it accurate; you set the standard.
Semantic contextdefinitions, exceptionsSkillshow the work gets done
OpenAIPlan → Query → Answerseconds, every questionEvalswhat keeps it honestThe Outer Loop
The inner loop is easy. The outer loop is the job.
Any harness can plan, query, and answer. Whether it's right depends on the context and evals around it.
What each topic and metric means.
Changes arrive as pull requests.
The numbers leadership believes.
How work gets done in your stack.
Graded questions on every change.
Running the outer loop
Be the center of gravity of your company's agentic workflow.
Drafted at scale, curated by you
We draft metric definitions, dbt changes, and business context from your stack. You curate the draft.
The loop learns from every correction
Thumbs-downs and new questions come back as proposed context and regression tests, and you review them before they join.
Evals on every change
Every pull request reruns the evals it touches and names the blast radius. Nothing merges until they pass.
feat: refund reason on int_orders_with_refunds #212
signalpilotproposed · feat/refund-reason → main
- add refund_reason to int_orders_with_refunds
- fix: keep 1:1 grain on order_id · signalpilot
- ✓Dana R. approved these changes
What changes in your week
Less labeling. More curating.
Curate, don't label
Questions stop landing on your desk as SQL requests. You review the proposed changes and curate.
Read the eval report
Every pull request shows the graded questions it touched.
See the blast radius first
Know which dashboards a change affects before an exec does.
Stays in your control
Your repo, your stack, your call.
See which models have evals, and which are passing.
- Snowflake
- Databricks
- BigQuery
Postgres
DuckDB- Redshift
- dbt
- GitHub
- Claude Cowork
- Codex
Slack
And the harness you already use.
Open source under Apache 2.0, every benchmark transcript public, and definitions that leave with you.
FAQ
Questions data leads ask first.
Short answers. The long ones are a demo away.
Do we have to document everything first?
No. We draft from your dbt models, dashboards, and agent traces, and you settle the conflicts.
We already have a semantic layer.
Keep it. We read the dbt Semantic Layer, Cube, or LookML and write Apache Ossie beside it.
How much of our time does this take?
An hour on day one, curation on day three, then a weekly pass over proposals, or hand it to our managed offering.