← Guides

Which AI tools automate GMP batch record review and release?

By Savi C., Co-Founder · Published Nov 17, 2026

Tool types and what each automates

Type of tool

AI review layer over paper and electronic records

What it automates

Reads executed records, including handwriting and scans, checks every entry against the master record, specifications and release criteria, and produces an exception list with each flag linked to its source page

Type of tool

EBR or MES with review by exception

What it automates

Enforces limits, sequence and signatures during execution, then shows QA only the steps and values the system marked as exceptions

Type of tool

QMS assistant

What it automates

Drafts or summarizes deviation, CAPA and change records inside that QMS, and can show which open records are linked to a batch

Type of tool

LIMS

What it automates

Not usually sold as AI, but supplies the test results and specification checks that release depends on

How to choose between them

Start with where your batch records live. If they are on paper, an EBR or MES only helps after a migration that rebuilds every master record in the system. An AI review layer can work on the paper you already have. Our guide AI record review vs an EBR or MES: do you need both? covers this trade in detail.

If you already run an EBR or MES, its review by exception covers what was captured in the system. It does not cover attached paper, printouts from standalone equipment, supplier certificates or checks across documents. That gap is where an AI review layer adds value. Our guide What is review by exception for batch records? explains how the method works and what QA still reviews.

A QMS assistant is a different job. It helps with the deviation and CAPA records that release depends on, not with checking the batch record itself. Many sites use one alongside the others.

What must stay human

Decisions

  • Approving production and control records before release, by the quality control unit (21 CFR 211.192 and 211.22)
  • Certification of each batch by the Qualified Person in the EU (Annex 16)
  • Deciding whether a discrepancy is explained or needs a full investigation
  • Accepting or rejecting each AI flag, with the decision recorded

Controls around the tool

  • Defining the checks the tool performs and the specifications it uses
  • Testing the tool for its intended use and controlling changes to it, as 21 CFR 211.68 expects for automated equipment
  • Periodic sampling of records the tool passed, to confirm it is not missing errors

How Litewave fits

Litewave is an AI review layer. It reads handwritten and scanned paper records, electronic batch records and system exports, checks executed batch records against your master record, specifications and release criteria, and produces a review-by-exception file for QA with every flag linked to its source page. It works on top of your QMS, MES, LIMS and ERP rather than replacing them.

Litewave proposes, qualified people approve. It never releases a batch on its own. At a leading global generic API manufacturer, batch disposition became 85% faster, as described in our published case study. For what inspectors and regulators look for when AI is part of review, see our guide Can AI be used for GMP batch record review? What regulators expect.

Questions we get asked

Can AI release a GMP batch?

No. In the US the quality control unit approves the records before release, and in the EU a Qualified Person certifies the batch. AI can check the record and present the exceptions. A person makes and signs the decision.

Do we need an EBR before we can use AI for batch record review?

No. An AI review layer can read paper batch records directly. An EBR helps prevent errors at entry, which is a separate benefit.

What does review by exception mean for an AI tool?

The tool checks every entry and shows QA only what failed a check or could not be confirmed, with the evidence for each. QA reviews those exceptions and the summary of what was checked, instead of reading every page.

Is AI-assisted review acceptable to FDA?

FDA requirements focus on the quality unit's review and approval, data integrity and control of automated systems. Nothing in those rules prevents software from supporting the review, as long as the tool is checked for its intended use, as 21 CFR 211.68 expects of automated systems, and people make the decisions.

Sources

  1. 21 CFR 211.192, Production record review (eCFR)
  2. 21 CFR 211.188, Batch production and control records (eCFR)
  3. 21 CFR 211.22, Responsibilities of quality control unit (eCFR)
  4. 21 CFR 211.68, Automatic, mechanical and electronic equipment (eCFR)
  5. EudraLex Volume 4, Annex 16, Certification by a Qualified Person and Batch Release
  6. Litewave case study, batch disposition at a leading global generic API manufacturer