PricingPay per right answer
A small seat fee, then $0.50 per successful thread and $0.01 per governed query. Model tokens at cost.
Team
Up to 5 seats and 200 data models
$100
a month · then $15 a seat
On us$50 of usage every month
- 5 seats included
- 30 data models with evals included
- Integration priced separately, below
From $2,000 a month
Scale
Most popularMore than 5 seats, or 200+ data models
$250
a month · then $15 a seat
On us$100 of usage every month
- 15 seats included
- 50 data models with evals included
- One integration week on annual
From $4,000 a month
Enterprise
Priced to your needs
Custom
a month · one purchase order
On us$1,000 of usage every month
- 100 seats included
- 100 data models with evals included
- Two integration weeks on annual
- SSO, SCIM, PrivateLink, customer-managed keys, audit export, 99.9% SLA
From $8,000 a month · named engineer
The same on every plan, on top of the plan fee. Failed threads are never billed.
Integration & custom data work
Our engineers on your schema, under NDA
Quoted per engineer-week
Scope and price agreed before any work starts
Includedone week on annual Scale, two on annual Enterprise
From $2,000 a month · priced on each plan above
A SignalPilot engineer keeps your evals and metrics correct every week, under the 90% precision guarantee.
- The integration week: dbt project fixed, evals written, knowledge base seeded, nightly run wired, agent live for one team.
- Custom data work on the same definitions: new marts, migrations, dashboards rebuilt, a second team onboarded.
- Delivered by SignalPilot engineers or a certified partner, on your schedule.
A team of 50 on Scale
50 seats, 200 dbt models with evals, and 1,000 completed runs a month.
The calculator adds queries, eval runs, Managed, model class, and the payback.
* All prices here are for illustration only. Your bill follows the rate card and your usage.
Every plan includes the governed gateway, verifier agents, an evidence graph on every answer, and 30 scheduled eval runs a month.
The guarantee
Credited the moment we did not deliver.
A thread that fails, or ran fewer than an SLA-backed percentage of evidence-backed queries, is credited without you asking.
90% precision on the graded questions we froze together, trailing 30 days. Below that, a credit on the month's outcome fees, down to a full refund under 80%. Model tokens are never refunded.
FAQ
The fine print, in plain words.
Short answers to the questions procurement asks.
Is billing annual?
Yes. A three-month plan is available at a 20% premium.
Why not just per seat?
Value is created when a question gets a correct, cited answer and when a data model stays correct, not when someone logs in. Seats are cheap so everyone who needs a number can have one.
What counts as a successful thread?
A thread that ended in a receipted answer: a governed query ran on your warehouse and you got the result with its evidence. Failures, clarifying questions, your data team's own tuning threads, repeats within 24 hours, and answers you flag as wrong within seven days are never billed.
What counts as a data model?
A named dbt model in your project: a mart, a fact or dimension table, a metric model. During the integration week we integrate the models that matter for your first workflow, and we co-author the metrics and definitions on top of them with your data team, so every covered model has an approved meaning and evals behind it.
What is an integration week?
A SignalPilot engineer on your schema: the dbt project fixed, evals written, the knowledge base seeded, the nightly run wired. Teams whose data estate fits their company size are usually integrated in one to two weeks. Larger or messier estates can take longer, and we scope that as custom data work. Quoted per engineer-week on Team; included on annual Scale and Enterprise.See the integration week