Why batch release is a software problem now
Batch release is slow for a simple reason: the evidence for one decision lives in many systems. The executed batch record is in an MES or on paper. Test results are in a LIMS. Deviations and change controls are in a QMS. Environmental monitoring, equipment logbooks and training records may sit somewhere else again. A reviewer spends much of the release cycle collecting and cross-checking, not judging.
Two trends are pushing companies to automate. First, regulators expect more checks before release, not fewer. The EU’s draft revised Annex 11 (consulted on from July to October 2025) says audit trail review “should be conducted prior to batch release” unless a later review is justified, and that release-relevant reviews “should be available to the QP at the time of batch release.” Second, FDA’s final Computer Software Assurance (CSA) guidance, issued in September 2025, makes validating production and quality software less burdensome by focusing effort on high-risk functions. More checks, with lighter validation, favors automation.
What batch release and batch record review software does
The market uses overlapping labels. It helps to sort tools by function:
| Category | Main job | Role in release |
|---|---|---|
| MES with electronic batch record | Runs and records manufacturing | Prevents errors at entry; produces exception reports |
| LIMS | Manages samples and test results | Supplies QC results and OOS status |
| eQMS | Manages deviations, CAPA, change control | Supplies open-item status for the batch |
| Release orchestration or “release cockpit” | Pulls status from the systems above into one view | Shows what is complete and what blocks release |
| Document review and assessment tools | Read records and documents and check them against rules | Find gaps and inconsistencies across records, including paper and PDF |
Most companies need more than one. A plant with a modern MES may still need a cross-system view. A plant with paper records may get more value from document review tools before investing in an MES.
What to automate, and what not to
Automation works best on tasks that are rule-based, repetitive and checkable. It works worst where the task is a judgment.
Good candidates for automation
- Completeness checks: every field filled, every signature present
- Limit checks: recorded values against specifications and master record limits
- Calculation checks: yields, reconciliations, potency adjustments
- Cross-record matching: material lots in the batch record versus released lots in inventory; equipment IDs versus cleaning and calibration status
- Status gathering: open deviations, OOS results, change controls and their states
- Audit trail filtering: surfacing changes to critical data for review
- Release checklist assembly: one record of what was checked, by whom, when
Keep with people
- Impact assessment of a deviation on the batch
- Deciding whether an investigation is sufficient
- Deciding whether related batches are affected
- The release or rejection decision itself
US regulations give the authority to approve or reject drug products to the quality control unit (21 CFR 211.22(a)). In the EU, certification is the QP’s personal responsibility under Annex 16. Software can prepare and present the evidence. It does not hold the authority.
Required controls for release software
If the software creates, changes or stores GMP records, or is used to make GMP decisions, it is in scope for electronic record rules.
US: 21 CFR Part 11
- Validation that shows “accuracy, reliability, consistent intended performance” (11.10(a))
- Secure, computer-generated, time-stamped audit trails where “record changes shall not obscure previously recorded information” (11.10(e))
- Authority checks so only authorized people can sign or change records (11.10(g))
- Electronic signatures that show name, date and time, and meaning, such as review or approval (11.50)
EU: Annex 11 (current version, with the 2025 draft adding detail)
- Risk-based validation across the lifecycle
- Audit trail review that is targeted and risk-based; draft section 12.5 says “the use of tools to help conduct audit trail reviews is encouraged”
- Independent review of audit trails by people not involved in the activity (draft 12.6)
Data integrity
FDA’s data integrity Q&A states that “when generated to satisfy a CGMP requirement, all data become a CGMP record.” If the software computes a yield or flags a deviation, that output is part of the record and needs the same controls.
Validating release automation with CSA
FDA’s CSA guidance applies to software used in production and quality systems. Its approach suits release software well:
- Define intended use. For example: “identify missing signatures and out-of-limit values in executed batch records before QA review.”
- Rate the risk. A function that blocks release or hides a finding is high risk. A dashboard that only displays status is lower.
- Match testing to risk. High-risk functions get scripted tests, including negative tests (does it catch a missing signature?). Lower-risk functions can rely on unscripted or vendor testing.
- Keep records proportional. Document the risk rationale, the tests run and the results.
For any function that lets reviewers skip parts of a record, the validation must prove that nothing important can be missed. Our article on review by exception explains how to justify that.
Questions to ask vendors
- Which of our systems can you read from, and how is the data transferred and verified?
- Can you handle paper or scanned records, or only native electronic data?
- How are your checks configured, and who controls the rules under change control?
- What happens when a check cannot run (missing data, failed interface)? Does it fail safe?
- How do you show the source for each finding?
- What Part 11 and Annex 11 controls are built in? Can we see your audit trail?
- What validation package do you provide, and how does it align with CSA?
- How do you handle software updates without invalidating our validation?
- If AI is used, is the model static and deterministic, and how are outputs verified by people?
Measuring the result
Track a few numbers before and after:
- Release cycle time: from batch completion to disposition
- Review effort: reviewer hours per batch
- Right-first-time rate: batches with no review findings
- Findings missed: issues found after release, at audit or in inspection, that the review should have caught
The last one matters most. Faster release with more missed findings is not an improvement.
Release software selection checklist
- Intended use written for each function
- Systems and record types in scope listed (including paper and PDF)
- Functions split into automated checks and human decisions
- Part 11 and Annex 11 controls confirmed
- Every finding traceable to its source record
- Fail-safe behavior when data are missing
- CSA-based validation plan with negative tests for high-risk functions
- Rules and configuration under change control
- Baseline metrics captured before go-live
See release automation on your own records
AskDona is a document and evidence assessment tool. It reads batch records, certificates of analysis, deviation reports and SOPs, including scanned paper, and checks them against 21 CFR 211, Part 11, EU GMP and your own procedures. Each finding links to the page that supports it, so reviewers verify rather than search. Run a sample batch release assessment on a recent batch package. AskDona shows the evidence; your quality unit decides.
FAQ
What is batch release software? Software that collects and checks the evidence needed for a batch release decision, such as batch records, test results and deviation status, and records the decision.
Can batch release be fully automated? The checks can be largely automated. The decision cannot: it belongs to the quality unit in the US and the QP in the EU.
Does batch record review software need validation? Yes, if it is used for GMP records or decisions. FDA’s CSA guidance allows a risk-based approach.
Is an MES enough for batch release automation? Often not. An MES covers the manufacturing record, but release also depends on LIMS, QMS and other records.
Can automation work with paper batch records? Yes, if the tool can read scanned documents. Document review tools can apply completeness and limit checks to paper records before a full EBR is in place.
References
All pages opened and checked by October 14, 2026.
| Source | Type | Used for |
|---|---|---|
| 21 CFR Part 11 (eCFR) | US regulation | 11.10 and 11.50 controls |
| 21 CFR Part 211 (eCFR) | US regulation | 211.22(a) quality unit authority |
| EU GMP Annex 16 | EU GMP guideline (copy hosted by ECA) | QP responsibility |
| Draft revised Annex 11 (2025 consultation) | EU draft guideline | Section 12 audit trail review |
| EC consultation page | EU consultation page | Consultation dates |
| RQA, FDA issues final guidance on Computer Software Assurance | Professional association news | CSA final guidance |
| FDA, Data Integrity and Compliance With Drug CGMP: Q&A | FDA guidance | All data become CGMP records |