Change control impact assessment that lists every affected SOP, system and filing before the board meets
Every affected SOP, system and filing listed before the board meets.
See one case, screen by screen ↓It is the step in GMP change control where QA and the reviewers work out everything a proposed change touches — SOPs, validated systems, equipment, specifications, products, market filings, stability, suppliers and training — before the change board decides. Change Control Impact Assessment has seven agents do that search, cite every item and add what similar past changes needed, for QA and the board to decide.
How to choose AI solutions for life sciences →
Updated
A change owner raises a change control in the QMS and writes what they think it touches. Then QA, Validation, QC and Regulatory CMC each go looking for the rest by hand: which SOPs, work instructions and master batch records name the part, which ERP, MES and LIMS records hard-code it, which equipment was qualified with it, which specifications and stability commitments it triggers, and what each market's registration says about it.
Something is usually found late. An incoming-inspection SOP that only knows one supplier's label, an MES recipe step that rejects the new barcode, a filing in a market nobody thought of — found after approval, after the first batch, or at the next renewal, as a deviation. ICH Q12 is adopted unevenly, so the same change can need no filing in one market and a prior-approval variation in another.
Estimated split for one Major change assessed by hand across four to six reviewers, and with the solution.
Seven agents grow each change into a cited impact tree and add what similar past changes needed; QA accepts, the change board signs.
Northwind Bio's Plant 3 in Leiden needs a second 2,000 L bioreactor bag supplier for Veltrimab and Corventa drug substance after its current supplier, Arvion Bioprocess, cut 2027 allocation by 40%. Here is change CC-26-1184, screen by screen, in the working solution.
Omar Haddad raised CC-26-1184 on Oct 2: qualify Tessmer Single-Use as a second 2,000 L bioreactor bag supplier, a Major supplier / material change on the agenda for Thursday Oct 8. Section 4 of his request lists the known impact: SOP-USP-031 and SOP-USP-044 to be revised, and a CBE-30 for Veltrimab in the US. The tree shows exactly that — three items.
“Section 4 · Known impact: SOP-USP-031 and SOP-USP-044 to be revised. Regulatory: CBE-30 for Veltrimab in the US.”
The Change Reader turns the request into what changes, where and for which products. Then four agents search in parallel: 2,412 SOPs and forms, 318 validated systems and 604 equipment items, 1,904 specifications, and 412 registration dossiers across 38 markets. Branches appear as they work — batch records, the MES recipe, the bioreactor qualification, the market filings.
Each branch is coloured by where it came from: listed by the owner, found in the sources, added from past changes, or checked and not affected.
The tree grows to 8 documents, 3 systems, 3 equipment items, 1 specification, 2 products, 10 market filings, 1 stability commitment, 2 supplier items and 1 training curriculum for 56 people. Twenty-six were found in the sources, and three were checked and found not affected — the drug substance release specifications, Talmirex, and Corventa in the US and EU.
“2 filings the owner did not list — no document names them, but similar past changes needed them: Veltrimab · JP and Corventa · UK.”
The comparer ranks 6,830 closed changes and reads the top three. CC-24-0611, a second sterilising filter supplier for Veltrimab (81% similar), needed a Japan minor change notification: Japan's approval matters name the bag material — “polyethylene-based bag” — not the supplier, so a supplier search misses it. CC-25-0219 (78%) missed a UK Type IB for Corventa, because the UK did not adopt Corventa's lifecycle document; it was filed four months late with a deviation.
The Past-Change Comparer adds a branch only with the past change that proves it, and never removes one.
Every branch opens the passage that proves it. MES recipe VLT-DS-USP v12, step 4.2, accepts only the current supplier's barcode format, so it would reject Tessmer's GS1 labels. At 78% confidence it is marked “your call”. The Cork precedent, CC-25-0732 (94% similar), found this rule and the incoming-inspection SOP only after approval, and its first batch was 9 days late.
“Step 4.2 rejects the new supplier's GS1 labels. The scope of the regression test is a judgement for Validation.”
The matrix lays the same items out by area, each with its disposition — Revise, Follow, Update, OQ addendum, New recipe — its confidence and its source: 12,480 sources searched, 43 passages cited. Dana accepts each item or marks it not affected. Three need a person's call because they sit below 85%: the MES recipe, Japan, and the TF-90 film's extractables data.
Items under the confidence line are never accepted in bulk.
The Registration Checker drafts 24 cells for Veltrimab and Corventa drug substance: 7 notify and wait, 7 implement then tell, 2 prior approval in Brazil, 2 no filing and 1 your call in Japan; the other 5 are markets where the product is not registered. Corventa needs no filing in the US or EU, because its approved lifecycle document (ICH Q12) says the bag supplier is not an established condition. The grid goes to Lena to confirm, and every cell stays a draft until she does.
“Release gate: no Tessmer-bag Veltrimab lot ships to the US until 30 days after FDA receives the CBE-30, or to Brazil before approval.”
Omar agrees. Lena agrees with a shipping restriction per market until the filing clock allows. Sam agrees with a risk-based MES regression test of the bag-lot step and the batch record print. Priya Raman agrees on condition that extractables coverage of the 30-day harvest hold is confirmed before the first GMP batch — the supplier's study covers 40 °C for 21 days. Quorum for a Major change, QA, Regulatory CMC and Validation, is met.
Dana signs with her password and the meaning “Approved as change board chair”. The decision carries three conditions: a shipping restriction per market, extractables coverage of the 30-day harvest hold, and no first batch until all 11 gating tasks are done. The three open calls go with it, each to its owner, to decide before the first GMP batch: the MES recipe to Sam Patel, Japan to Lena Ortiz, the TF-90 extractables data to Priya Raman.
“Your signature applies to CC-26-1184 — approve with 3 conditions and 3 open calls carried as conditions.”
The planner turns accepted items into 14 tasks with owners and dates, each linked to the branch that created it. Eleven gate the first batch: documents, the raw material specification, ERP, MES and LIMS, the OQ addendum for BR-301 and BR-302, the Tessmer audit and quality agreement, the extractables check and training for 56 people. The first batch on Tessmer bags can be released on Nov 20, six weeks after approval; the Cork precedent took seven.
The release gate always includes supplier qualification and training.
The same change, seen by the five people who assess it — in their own words, and what their week looks like now.
Seven agents grow each change into a cited impact tree and add what similar past changes needed; QA accepts, the change board signs.
Reads the change request and turns it into what changes, where, for which products and materials, and what the owner already named as impact.
Searches SOPs, work instructions, forms and master batch records for every document the change makes wrong, and the people trained on each.
Matches the change to validated computerised systems (ERP, MES, LIMS) and qualified equipment, with a proposed assurance level.
Checks specifications, methods, products and stability commitments — including the ones that are not affected, with the reason.
Reads each registration and the country rule books and drafts the filing category per product and market, through an ICH Q12 lens.
Finds the most similar closed changes and adds the branches they needed that this change does not yet have — the missed-impact catch.
Writes the cited impact assessment for the board, and after approval turns accepted items into an implementation plan with a release gate.
Votes and signs, with conditions. The agents propose; a named person decides.
Ask about any change, its impact, the filings or the board — every answer points to the passage behind it.
What did the owner of CC-26-1184 miss?
CC-26-1184 touches 31 items in 9 areas; the owner listed 3. Two of the others are filings that only past changes revealed: Veltrimab in Japan, where approval matters name the bag material, not the supplier, and CC-24-0611 needed a filing for the same reason; and Corventa in the UK, which did not adopt the lifecycle document — CC-25-0219 filed four months late. Also likely to stop the first batch: incoming inspection SOP-WH-012 and the MES label rule.
Which markets need a filing for CC-26-1184?
Draft filing plan, for Lena Ortiz to confirm. Veltrimab: notify and wait in the US (CBE-30), EU, UK and Switzerland (Type IB), Canada and Singapore; prior approval in Brazil; implement then tell in Australia, China and Korea; your call in Japan. Corventa: Type IB in the UK; no filing in the US and EU, because the bag supplier is not an established condition. No Tessmer-bag Veltrimab lot ships to the US until 30 days after the CBE-30 is received.
Is the E&L data enough?
Probably, but it is QC's call. The supplier's study covers 40 °C for 21 days; the culture runs 16 days at 36.5 °C, so that is covered. The harvest can sit in the bag for up to 30 days at 2–8 °C — colder, but longer than 21 days. SOP-QC-207 asks for worst-case conditions. Confidence 62%, so it is in Priya Raman's queue as a condition before the first GMP batch.
Show me changes waiting on Validation
Three changes are waiting on Sam Patel: CC-26-1184 for the MES regression scope and the OQ addendum, CC-26-1175 for the EM software assurance plan, and CC-26-1171 for the interface retest, 70% done.
13 screens from the working solution, on its sample data. Pick one to see it large.
How the agents assess changes and how the board decides — set once, in plain words.
| Setting | Default | Choose from |
|---|---|---|
| Send to a person below this confidence | 85% | 75% · 85% · 90% |
| Filing categories are drafts | On — Regulatory CMC confirms every product × market category | Required by policy |
| Look for precedents at every site | On — search 6,830 prior changes for one of the same kind | On · Off |
| Assess new changes as soon as they arrive | On | On · Off (the owner starts the assessment) |
| Board meets | Thursdays · 10:00 | Agenda pack goes out the afternoon before |
| Major changes need | QA, Regulatory CMC and Validation | QA · Regulatory CMC · Validation · QC |
| Minor changes approved without the board by | QA alone | Who can approve without the board |
| Rules in plain words | “A change to a product-contact material always opens an extractables and leachables item.” | Add a rule — added by Priya Raman, Aug 2026 |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample change controls · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (ISPE and PDA on ICH Q12 — global approval of a simple post-approval CMC change can take up to five years). People, companies, products, suppliers and change records named on this page are fictional — characters and sample data in the working solution.
It is the step in GMP change control where QA and the reviewers work out everything a proposed change touches — SOPs, validated systems, equipment, specifications, products, market filings, stability, suppliers and training — before the change board decides. Change Control Impact Assessment has seven agents do that search, cite every item and add what similar past changes needed, for QA and the board to decide.
Four agents search the document library, validated-system and equipment inventories, specifications and registrations in parallel. Then the Past-Change Comparer ranks closed changes by similarity, reads what the top three needed — including items found late or after implementation — and adds any branch this change lacks, in amber, citing the past change.
The Registration Checker reads each registration, its lifecycle document where there is one, and the country rule book, and drafts a category per product and market — prior approval, notify and wait, implement then tell, or no filing. Every category is a draft until Regulatory CMC confirms it, and below 80% confidence it is marked “your call”.
No. The agents propose a disposition such as Revise, Follow, Update or Not affected, with the passage behind it and a confidence. QA accepts each item or marks it not affected, and items below the confidence line go to a named person and are never accepted in bulk.
No. The agents read the QMS, document library, RIM and inventories, and never write to them without a person. After the board signs, the planner creates tasks only from accepted items.
Yes. QA accepts each branch, Regulatory CMC confirms the filings, Validation sets the assurance scope, the change board votes and the chair signs with the meaning of the signature. Open calls become conditions with an owner, and no changed batch is released until every gating task is done.
Yes. The confidence line, precedent search, assessing on arrival, board day and time, the votes a Major change needs, who approves Minor changes and plain-language rules are all settings. Draft filing categories and electronic signatures are required by policy.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your change control SOP, document library, system and equipment inventories, registrations and a sample of closed changes — and it goes live once every quality gate has passed. We will walk you through it on your own changes first.
Any batch-record question answered in seconds, cited to the page and step.
1-min demoValidation Document AutomationProtocols drafted in minutes, every value traced to its source.
1-min demoManufacturing Batch AnalyticsBatch analysis that took three months now takes hours, with the query behind every answer.
1-min demoWe'll run Change Control Impact Assessment on a sample of your own closed change controls.