Approve the data boundary
Agree one site, one CAPA/NCR workflow, usable fields, retention, access and the authoritative customer record.
LEFTFOUNDING MANUFACTURER PROGRAMMEProof Sprint · £2,950–£4,500
HOW IT WORKS
Start with approved quality history.
Keep every decision with your team.
RepeatProof adds a secure, fixed-scope search and review layer around your existing records. It finds and prepares evidence; your QMS remains the official record and your authorised people make the decisions.
CHECK · 30 MINUTESFree Quality Memory AssessmentPROVE · 4–6 WEEKSFounding Manufacturer Proof SprintFOUNDING PRICE£2,950–£4,500OPERATEOngoing use after measured proof
FIVE CONTROLLED STEPS
From approved history
to a measured decision.
Every step has a named owner, evidence requirement and review point. Progress never implies approval to integrate, automate or scale.
Build the quality memory
Prepare approximately 250–1,000 approved historical records so different language and structures can be searched consistently.
Find related history
Surface potentially related events, previous actions and available effectiveness evidence with the reasons each record appeared.
Run named human review
Authorised reviewers decide which AI-surfaced evidence to include, investigate further or exclude after inspecting sources, relationship scores, limitations and evidence gaps.
Measure and decide
Compare preparation time, precision, recall, Top-K recall, false-negative rate and reviewer agreement before a proceed, change-scope or stop recommendation.
WHAT REPEATPROOF NEVER DOES
Evidence support
without quality authority.
The boundary is simple enough to explain before data exchange and visible throughout the review.
No root-cause or relevance decision
Similarity is a retrieval aid. The authorised investigator decides what is relevant and what the evidence means.
No disposition, release or closure
RepeatProof cannot approve product disposition, corrective-action effectiveness, product release or CAPA closure.
No replacement QMS
The official quality record stays in the customer's controlled system and process.
WHAT GETS MEASURED
Agree the method
before seeing the result.
Recurrence or scrap reduction may require a longer observation period. The Proof Sprint first tests whether retrieval and evidence preparation improve on an agreed labelled benchmark or known-relevant set.
Preparation time
Comparable evidence-pack preparation using agreed start and stop points.
Precision
Reviewer-accepted related records divided by all displayed records.
Recall
Known relevant records retrieved divided by the agreed known-relevant set.
Top-K recall
Recall at the first 5, 10 and 20 results shows whether useful evidence appears early.
False-negative rate
Known relevant records missed divided by all known relevant records.
Reviewer agreement
Comparable reviewer decisions that agree, with disagreement reasons retained for review.
WHAT THE CUSTOMER PROVIDES
Named roles and time.
No hidden commitment.
These are planning ranges for the default Proof Sprint. The proposal confirms people, records, review time, meetings and decision owners before work begins.
Bounded customer participation
- Quality lead: Approximately 4–6 hours across the 4–6 week sprint
- Data owner: Approximately 2–4 hours to export, map and confirm the approved history
- Reviewers: Two or three named reviewers, normally 2–4 hours each for structured record review
Approved history with usable comparison fields
One site, one CAPA/NCR workflow and approximately 250–1,000 usable historical records.
- Event references, types and dates
- Problem or failure-mode context
- Action links and effectiveness evidence where available
Three planned touchpoints
45-minute kick-off, 30-minute progress review and 60-minute closeout. Up to 60 minutes from security or procurement when buyer due diligence is required.
What can change the effort
Planning ranges are confirmed in the proposal. Data remediation, extra sites, live integration and validation work are separately scoped.
Check the fit
before committing to delivery.
The free 30-minute Quality Memory Assessment ends with Fit, Possible Fit or Not Suitable Yet. A Founding Manufacturer Proof Sprint is proposed only when a credible scope can be agreed, and production follows only after measured proof.