Deviation and CAPA management with the evidence for and against every root cause
Every deviation closed inside the 30-day target, with a CAPA aimed at the cause.
See one case, screen by screen ↓When a filter integrity test fails, a fill weight drifts or a cold room warms, QA has to investigate, find the root cause, assess product impact and raise CAPAs — usually against a 30-day target. Most of that time goes on collecting evidence: the batch record, equipment run logs, maintenance work orders and plans, environmental and utility data, and past deviations, each in its own system. The investigator pulls them by hand while the batch waits on hold.
Under that pressure the investigation settles on the first plausible cause. Root causes default to human error and CAPAs to retraining, and a CAPA is scoped to the line that failed, not to the same equipment on the other lines. So the same failure comes back. 21 CFR 211.192 asks for a thorough investigation of any unexplained discrepancy, extended to other batches that may be affected.
Estimated split for a typical Minor or Major deviation, in calendar days, by hand and with the solution.
Five specialist agents fill the fishbone from the records, weigh each cause for and against and tie the CAPA to it; QA approves and signs.
Plant 3 in Leiden fills Corventa 20 mg vials on aseptic filling Line 1. On Oct 6, 2026 at the 09:50 in-process check, nozzle 4 starts filling light. Here is what happens next, screen by screen, in the working solution.
Dana approves root causes and CAPAs for Plant 3. Her board shows 10 open investigations, each on its 30-day clock, and one of four waiting for her approval: DV-2026-0412, a filter integrity test failure on Line 2. Its six bones already carry what the agents found, and the why chain runs from the failed test to the system cause — CAPA-2025-0311 changed the Line 1 batch record only.
Proposed root cause: “Filter tested while still warm from the hot water flush” — Method, 82 % · QA decides
At the in-process weight check, 4 of 5 vials from filling nozzle 4 weigh 1.95–1.98 g against a limit of 2.00–2.10 g; the other seven nozzles are in range. The line stops at 09:54. Sam reports it in one sentence, with the area and the time, against batch CRV-24-0546.
Clock keeper: “Minor — every light vial rejected by 100 % weighing” · clock started, due Nov 5.
The agents read the records for the event window, bone by bone. Method: the recipe audit trail shows CRV-20-V3 verified at 06:12, 2.02 mL on all 8 pumps. Measurement: the check-weigher's daily check passed. Environment: bulk solution at 20.6–20.9 °C, no alarms. Material and People come back clear. Then the machine bone fills from the maintenance records.
“Evidence gatherer · Filler manual FM-L1 · section 9.2: seal change due at 1,500 h”
Work order WO-2026-3318: the nozzle set has run 1,840 operating hours. Filler manual FM-L1 §9.2.3: replace the tip seals every 1,500 hours. PM-L1-FIL-07: the Sep 29 nozzle inspection was deferred to Oct 14. The analyst proposes a worn tip seal on nozzle 4 at 76 %, and lists the evidence against: no drip-back seen yet — engineering inspects at 14:00 — and pump 4 could give the same signal.
Why chain: “Why not changed? The plan counts months, not hours, and the Sep 29 inspection was deferred.”
Replace all 8 tip seals on Line 1, because every seal has the same 1,840 hours. Change tip seals by operating hours on Lines 1, 2 and 4, as Line 3 already does. Deferring a nozzle inspection needs QA review. The check-weigher alerts when one nozzle drifts 1 % from the others. Raising the nozzle 4 dose by 0.06 mL is shown struck out, with the reason.
Not proposed: “Moves the number back but leaves the worn seal in place. The drift would return, and the other seven seals have the same hours.”
In the investigation, every finding is a card with its source and a tag — key finding, contributing, pattern, ruled out, argues against pump. The check-weigher data shows only nozzle 4 drifting, from 2.05 g to 1.99 g since Sep 24, with 1,212 vials rejected low. The similarity finder flags DV-2025-0466: the same worn seal on Line 3, at 1,710 hours, and its CAPA-2025-0402 changed the Line 3 plan only.
“This is a repeat. Lines 1, 2 and 4 use the same nozzle set.”
Worn tip seal on nozzle 4 at 76 %. Piston wear on dosing pump 4 at 17 % — argued against by a gravimetric check that passed on Sep 15 at 2.021 mL and a piston replaced in January. Recipe dose not reloaded at 5 %, because a recipe error would move all nozzles, not one. Check-weigher reading low at 2 %, ruled out.
Against a recipe error: “A recipe error would move all nozzles, not one.”
The report writer drafts the investigation report in the plant template — summary, event and immediate actions, data reviewed, root cause, product impact and the CAPA plan — with a citation on every claim. Product impact: no light vial reached the batch, because the check-weigher weighs every vial and rejected all 1,212 below 2.00 g.
“Draft v1 · written by the report writer · every claim cited”
Dana reviews the root cause, the CAPA plan of 5 actions with its effectiveness check, and the product impact on one screen. She signs as QA approver with her user ID and password, choosing the meaning of her signature: “I approve the root cause and the CAPA plan”.
Agents propose; only named people move a deviation to CAPA or Closed.
DV-2026-0419 moves to CAPA, and four actions open as CAPA-2026-0094 to 0097: Omar Haddad owns the seal changes, Lena Ortiz the QA review of deferrals and Jonas Weber the drift alert. The effectiveness check is set: no single-nozzle drift above 1 % on Lines 1–4 in the next 90 days — review Jan 15, 2027.
“Signed — DV-2026-0419 moved to CAPA · 4 actions opened as CAPA-2026-0094 to 0097”
The same deviation, seen by the five people who own it — what their month looked like, and what it looks like now.
Five specialist agents fill the fishbone from the records, weigh each cause for and against and tie the CAPA to it; QA approves and signs.
Reads the deviation report and pulls the batch record steps, equipment run logs, utility logs and environmental data for 24 hours around the event, quoting each fact with its source passage.
Searches 36 months of deviations and CAPAs for the same failure mode, equipment class and product, scores the similarity and checks whether earlier CAPAs covered this line.
Fills the six bones — Method, Machine, Material, People, Measurement, Environment — ranks candidate causes with a confidence and the evidence for and against each, and builds the 5-Why down to a system cause.
Drafts the investigation report in the plant template — summary, event, immediate actions, data reviewed, root cause with 5-Why, product impact, CAPA plan and effectiveness check.
Proposes the class from the plant's rules at intake and starts the 30-day clock; counts the days of every open deviation each morning, warns the owner at day 20 and drafts an extension request at day 25.
Approves root cause and CAPA. The agents propose; a named person decides.
Ask about any deviation, root cause or CAPA — or say what to change. Every answer points to its records.
Why is nozzle 4 filling light?
Most likely a worn tip seal on nozzle 4 (76 %). The nozzle set has run 1,840 hours; the supplier says change the seals at 1,500, and the Sep 29 inspection was deferred. Only nozzle 4 drifts, a little each day since Sep 24. Line 3 had the same cause in 2025 and its fix stayed on Line 3. The CAPA changes seals by hours on Lines 1, 2 and 4 — not a dose adjustment.
Which investigations could miss the 30-day target?
Two are at risk: DV-2026-0352, visual inspection rejects on Line 2 — day 28 of 30, extension request drafted for Dr. Maya Chen; and DV-2026-0364, buffer pH in Suite 2 — day 26, with one CAPA due Oct 12 still in progress.
Why did the Line 2 filter test fail?
Most likely the filter was tested while still warm. It was flushed with water from the hot loop at 79.6 °C and tested 6 minutes later at 41.2 °C; the housing cooled 3.4 °C during the measurement, which reads as extra flow. The retest after a 30-minute cool-down passed at 9.8 mL/min. It is a repeat: the 2025 fix changed the Line 1 batch record only.
Make every sterile-filter test failure a Major deviation
Done. New classification rule: “Any failed integrity test of a sterilizing filter → Major, batch on hold.” It applies to deviations opened from now on; you can change it in Settings.
13 screens from the working solution, on its sample data. Pick one to see it large.
The clock, the approvers and how the agents work are settings, chosen by the plant.
| Setting | Default | Choose from |
|---|---|---|
| Target to close an investigation | 30 days | 30 days · 45 days · 60 days, in calendar days from opening |
| Warn the owner | At day 20 | At day 15 · At day 20 · At day 25 |
| Start the agents as soon as a deviation is opened | On | On · Off — when off, the investigator starts them from the workspace with one click |
| Root cause and CAPA plan approver | Dana Okafor | Dana Okafor · Dr. Maya Chen — electronic signature with its meaning recorded |
| CAPA effectiveness | Lena Ortiz | Lena Ortiz · Dana Okafor — rates each CAPA effective or not effective |
| Look back for similar deviations | 36 months | 24 months · 36 months · 60 months |
| Call a cause “most likely” only above | 50 % confidence | 50 % · 60 % · 70 % confidence |
| Check every CAPA's scope across all lines | On | On · Off — flags lines with the same equipment that an action does not cover |
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 plant data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (McKinsey & Company, Generative AI: a game changer for biopharma operations — about 40 % shorter deviation closure; 15–20 % of deviations recur). Regulatory context: 21 CFR 211.192; ICH Q10 §3.2.2. People, the plant, products, batches and records named on this page are fictional — characters and sample data in the working solution.
Software that takes a GMP deviation from the moment it is reported to a QA-approved root cause and a CAPA plan, with owners, due dates and effectiveness checks. Deviation & CAPA Management adds five agents that gather the records, fill the fishbone, weigh each cause for and against and draft the report — QA reviews and signs.
The evidence gatherer reads the batch record, equipment, maintenance, utility and room records for the event window. The root-cause analyst fills the six bones of the fishbone, ranks the candidate causes with a confidence and the evidence for and against each, and builds the 5-Why down to a system cause.
It is built not to. The root-cause analyst accepts human error only with the process reason behind it, and shows what argues against each cause as well as what supports it. Investigators and QA see the whole case before anyone signs.
The similarity finder searches past deviations and CAPAs for the same failure mode, equipment and product, and checks each earlier CAPA's scope across lines. In the sample case it found that the 2025 fix for the same worn seal changed the Line 3 plan only, so the new CAPA covers Lines 1, 2 and 4.
The clock keeper proposes the class from the plant's rules and starts the clock at intake, warns the owner at day 20 and drafts an extension request at day 25 with the reason, the work left and a new date. It cannot extend a clock; the Head of Quality decides.
Yes. The agents have read-only access to GMP records, propose and draft, and never approve. Root cause and CAPA need a QA electronic signature with its meaning recorded, and only named people move a deviation to CAPA or Closed.
Yes. The closure target, warning and extension days, approvers for root cause and CAPA, Critical deviations, extensions and effectiveness, whether the agents start on their own, the look-back for similar cases, the confidence threshold, the CAPA scope check and the classification rules are all settings.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your deviation and CAPA procedures, report template and a sample of past investigations — and it goes live once every quality gate has passed. We will walk you through it on your own deviations 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 Deviation & CAPA Management on a sample of your own closed investigations.