Trueness.ai anchor pilot: two quarters
25 October 2026. One $600,000 annual Foundation anchor plus two seven-cell Enterprise Audits at $105,000 each, over six operating months. These are planning contracts, not signed customers. The programme has three audiences in two markets. Audits may be in adjacent consumer categories but each has its own cell map, protocol and finding.
Entry and commercial readiness
Name the executive sponsor, research owner, commercial/legal party, programme lead, external reviewer and separate holdout custodian. Complete source/IP permissions before development or ingestion; general experience does not transfer proprietary research rights. Secure exact-scope quotes, recruitment feasibility, bilingual review where necessary and protected supplier slots. Use the enterprise contracting SOP for the MSA/SOW, data/processor/security schedules, agency access, payment timing and model-change obligations.
The six-month operating clock assumes the anchor signs in M1. Enterprise prospecting may take six-to-nine months and is not evidence of a signed pipeline. The model includes prelaunch costs and sales-delay sensitivity. Fieldwork is funded before net60 client receipts; signature alone does not fund recruitment.
Schedule and gates
| Period | Work | Evidence needed to proceed |
|---|---|---|
| Prelaunch | Quote actual cells; secure executive sponsor, source rights, named research/reviewer capacity and cash | Signed conditional scope and funding; no promise that every audience will pass |
| M1–M2 | Anchor contract; two-market workspace; first 72 qualitative interviews and 960 developmental completes; protocol/configuration freeze | Isolated clients/agencies, protected holdouts, stimulus completeness, signed method |
| M3 | First full wave: 1,440 sealed completes; six cell records; Panel/Verdict/Challenge/Talk/research/export; agency read-only MCP | Per-cell release/restriction decisions, current status in every surface, trained agency users |
| M3–M4 | Sign two seven-cell audits; freeze incumbent outputs and new stimuli; contextual interviews | Exact access/permissions, eligible participants, independent custody and capacity |
| M5–M6 | Deliver audits; second Foundation wave: 480 development +1,440 sealed; QBR and cost review | Fresh prospective comparisons, all failed/inconclusive endpoints retained |
| M6 | Paid continuation/channel-readiness gate | Useful repeated workflows, funded delivery and defensible claims; annual renewal remains future |
Two Foundation waves: 4,320 quantitative completes, 2,880 sealed and 72 interviews. Two audits: 3,360 sealed quantitative completes and 56 contextual interviews. Total: 7,680 quantitative completes, 6,240 sealed and 128 interviews, across separate populations/cells and dates. Do not pool these into a single validity sample. Repeat exposure or overlapping audience membership must be identified and handled in analysis.
An audit cell covers one registered task/instrument and up to four reserved concepts. Context interviews explain scope and gaps; they cannot repair the frozen system against its own final test. Expanded markets, six audiences or higher precision require separately costed work. The method defines limits and rotation.
Money and capacity
The six-month budget recognises $410,000 revenue from $810,000 first-year contracted value. Cash outlay is approximately $1,287,300 including prelaunch and half of the $750,000 first-year platform budget; receipts are $255,000 by M6. Funding including 20% contingency is approximately $1,238,760. An annual anchor is not recognised or collected in full at signature. This is the specified enterprise build, not a minimum-cost concierge option.
Reserve a whole programme research lead from signing, separate audit delivery hours and reviewer days, engineering/security capacity and agency enablement. The 0.5-FTE-per-Foundation ratio is a steady capacity target, not a reason to under-resource the first client. Research sign-off and custody cannot be delegated to model agreement. Replacement fieldwork and model-change overruns require a funded response.
Success, failure and continuation
Measure practical usefulness separately from calibration: recurring users across brand and agencies, time to usable response, panels followed by a documented change/export/human follow-up, research questions answered with correct citations, and decisions whose later outcome is recorded. Record the pre-pilot workflow and human time so savings have a denominator. An activity target is a hypothesis, not proof of better decisions or incremental revenue.
Every predictive claim needs its preregistered statistical release gate. A productive workspace with no supported quantitative claim can continue only as clearly bounded exploration/retrieval under an amended scope. High usage cannot override a failed claim. A statistically adequate but unused service also fails the commercial test.
At M6 decide: continue the annual service, narrow/remediate a scope, fund additional evidence, or stop expansion. Record whether audits lead to paid programme commitments; they do not automatically convert. An executed licence or qualified, dated opportunity is evidence; a vaguely “signed channel conversation” is not. A true annual renewal is measured at the service anniversary.
Build and launch backlog
- Legal entity, rights decision, enterprise contract schedules and supplier quotes.
- Programme/brand/agency hierarchy; protected evidence, holdout and record stores.
- Six everyday actions with scope labels, source completeness and recovery.
- Foundation first-wave workflow and independent record release.
- Read-only MCP for the anchor's agency by M3, with isolation/revocation tests.
- Second-wave rotation, model-version change workflow and QBR dashboard.
- Two audit deliveries and actual-cost reconciliation.
- Channel qualification after evidence, capacity and customer usefulness justify it.
No launch, customer contact, supplier purchase or production product implementation is represented as completed by this plan.