The programme
Day 0 → 30 — Atomic prototype
dbt manifest and schema ingestion engine; automated invariant generator; in-perimeter signed HTML deliverable. Gate: the generator synthesises at least 25 valid relational rules from a real manifest in under 15 minutes.
Day 30 → 60 — Two paid design-partner sprints
Two $15,000 Baseline Assurance Sprints, card-billed, delivered end to end. Gate: ICFY ≥ 40% over the native platform evaluation, and the CDAO — not us — validates that the defects are material.
Day 60 → 90 — Continuous conversion
The
evalqaCLI integrated into a partner’s weekly dbt CI. Gate (critical): at least one partner converts to the $48,000/year platform and authoring time is proven under 10 hours.
The kill rule
If neither of the first two design partners converts to the recurring platform because one-time audits are sufficient, we halt the platform build. We say this here so you know the offer is a test, not a pitch. Concretely: if by Day 90 neither design partner converts to the recurring platform because “one-time audits are sufficient”, platform software development halts and the company pivots to a lean security and audit advisory practice.
If at least one design partner converts and authoring time is under 10 hours, we hire the Senior Analytics Engineer and begin fundraising. Headcount is gated on contracted ARR and backlog, not on customer count.
The other throttle rules
The Day-90 gate is not the only one. Each signal below has a diagnostic and a structured action, so a bad result is not argued away.
| Observable signal | Diagnostic | Action |
|---|---|---|
| Native parity: ICFY < 20% across 5 customers | Measure ICFY against a locked customer baseline. | If native evaluation catches ≥80% of what we catch, halt standalone product investment; narrow to multi-role security audits. |
| Price resistance: prospects reject the $15,000 fee | Is friction driven by persona, packaging or unclear ROI? | Test a focused $10K scope; freeze hiring until two close. |
| Consequential irrelevance: Decision Change Rate < 20% | Do findings alter decisions? | Re-anchor test generation to high-exposure financial models. |
| Margin compression: sprint COGS > $12,000 or contribution margin < 50% | Labour hours per query across delivery logs. | Halt bespoke authoring sales; mandate automated synthesis before new engagements. |
| InfoSec blockers: security blocks > 30% of pipeline | Review redlines across stalled deals. | Mandate Mode 1 execution and the in-perimeter signed bundle; guarantee zero raw export. |
What a design partner gets, and gives
You get
- Founder-led delivery of both sprints at $15,000 each, card-billed
- First call on the invariant catalogue and the reporting format
- The invariant suite in your repo, yours to keep regardless of outcome
- A platform subscription at the published price if you convert — no founder discount, because a discount would contaminate the test
- The Common Paper Design Partner Agreement v1.3, unchanged, so counsel has nothing new to read
You give
- An agent in production or piloting against a real warehouse schema
- A named Customer Business Definition Owner with a 48-hour SLA on INDETERMINATE findings
- An honest Day-90 answer, including “one-time audits were enough”
- Permission to be named as a design partner, and a case study you approve line by line
The qualification bar
- A production or pilot deployment of a Snowflake Cortex Agent, Databricks Genie space or custom Text-to-SQL agent against a schema you can grant read-only metadata access to.
- A consequence tier above exploratory. The first cohort is deliberately Tier 1 or Tier 2 — financial, payroll, operational — because that is where ICFY is meaningful.
- The three roles in one room at kickoff: executive sponsor, technical lead, Business Definition Owner.
3 slots. We will say publicly how many remain. Today: 3.