01
Qualify
Sales qualification
Confirm the regulated use case, data constraint, accountable owner, urgency, and commercial fit.
→ Delivery
A gated route from restricted data to accepted evidence.
MirrorFoundry is sold and implemented as a high-assurance data capability—not activated from a generic registration flow.
03
SOW010203040506
Relational dataTime seriesEvent streamsFree textEdge-case simulationPrivacy evidenceRelational dataTime seriesEvent streamsFree textEdge-case simulationPrivacy evidence
Six gated handoffs
No data release or model decision without an owner.
Each gate creates a reviewable artifact, a decision, and an accountable next step.
02
Design
Privacy & model validation plan
Define source boundaries, intended use, threat model, utility tasks, simulations, and acceptance evidence.
→03
Agree
Compliant SOW
Document processing roles, scope, environments, retention, subprocessors, controls, acceptance criteria, and commercial terms.
→04
Provision
Private environment
Provision an isolated workspace and connect only the approved data sources, destinations, and model interfaces.
→05
Accept
Data & model acceptance
Run privacy, fidelity, task-utility, robustness, and operational tests against the signed protocol.
→06
Launch
Admin-led invitation
The customer administrator invites authorised employees after acceptance. No public signup is offered.
→What the SOW resolves
Scope the data, evidence, and responsibility together.
The SOW is the bridge between a promising synthetic-data use case and an accepted production capability.
01Intended use and exclusions
02Source and field boundary
03Privacy threat model
04Utility and model tasks
05Acceptance thresholds
06Deployment and residency
07Roles and responsibilities
08Retention and exit plan
Start with one release
Bring one constrained decision to qualification.
We’ll frame the source boundary, intended use, privacy risk, model evidence, and delivery conditions before proposing implementation.
Request an assessment