Eluate
Eluate mark: a developed TLC plate with three spots

Eluate

el·u·ate noun

  1. chromatography What comes out of the column: the clean, separated result.
  2. AI tools that take QC lab work to a result ready to evaluate.
AI tools for pharma QC

Less repetitive paperwork. Every rule auditable. Time back for the work that matters.

The same checks, comparisons and write-ups, done again and again, wear a QC team down, and that is where errors creep in. These tools take on the repetitive part so people spend their time on the calls that need them, and a tool stops rather than guesses when evidence is missing.

See the toolsRun a pilot
Prototypes on synthetic dataNot for GMP use
Tools

One tool per lab problem

Current examples, each filed below as a record: what it does and how it was tested.

EL-01script

Compendial Change Tracker

Flags what changed in USP-NF since the last check.

26 itemsflagged in one run, July 2026
Read the record →
EL-02agent

Deviation Review Agent

An AI second reviewer, a verifier, and a human sign-off that can’t be skipped.

15 of 15serious planted flaws caught
Read the record →
EL-03workflow

Reduced-Testing Writer

From an evidence pack to a reduced or ID-only testing justification.

27 of 27test runs matched the answer key
Read the record →

How the tools are built

The rules, the test sets and the answer keys come from QC practice, and every AI result is measured against them.

Rules first

What counts as a flaw, what blocks a reduced test, what the tool must refuse to do. Written as a spec before any code.

Known-answer test sets

Synthetic records with known flaws and known outcomes. The answer key is fixed in advance, so the model never grades itself.

Measured against the key

Every AI result on this page is scored against an answer key fixed before the run.

Each tool is labelled by how much it decides on its own: a script is fixed code, a workflow is fixed steps with AI at some of them, an agent chooses its next step.

EL-01 · script

Compendial Change Tracker

RecordEL-01
KindScript, no AI
DataPublic USP-NF pages
OutputExcel report of new items

USP-NF changes arrive as notices of intent to revise, revision bulletins and errata. Somebody in the lab has to spot the new ones and check whether any touch the lab’s products. This tool does the spotting: it lists only what is new since the last check, so the product check starts from a short list.

What it does

It reads the three public USP-NF update pages, compares every entry against a saved baseline, and writes an Excel report of what is new. It is deterministic Python with no model in it, because nothing here needs judgment: it is a list comparison.

How the Compendial Change Tracker works: three public USP-NF pages are parsed, compared with a saved baseline, and only new items go into an Excel report. No AI. 3 public USP-NF update pages saved as text Parse every entry type, title, date Compare with saved baseline list diff baseline.json Excel report new items only 26 on 6 Jul 2026 deterministic Python,no model anywhere
fixed codeHow a run works.
Notices of intent to reviseGeneral announcementsPublication announcementsGeneral chapter prospectuses
Sample output: July 2026 run
General AnnouncementCall for Data: Request for Stakeholder Information Related to Pharmacopeial Forum (PF) 52(2)26-Jun-2026
General AnnouncementProposal Status Reporting on USPNF.com26-Jun-2026
Notice of Intent to ReviseBemotrizinol09-Jun-2026
Notice of Intent to ReviseAlpha Lipoic Acid29-May-2026
Notice of Intent to ReviseErythromycin29-May-2026
Notice of Intent to ReviseMethyl Salicylate29-May-2026
+ 20 more items in the same run
The first 6 of the 26 new items, copied from the run file and set as a table here. That run had no new revision bulletins or errata.
EL-02 · agent

Deviation Review Agent

RecordEL-02
KindAgent + verifier + human gate
Data12 synthetic deviations
Human gateRequired

An AI second reviewer for deviation investigations: every finding quoted from the record, every finding challenged, and a human sign-off that can’t be skipped.

What it does

  • Reviewer agent reads one deviation investigation, the way a QA second reviewer would. It can read one file in an empty folder and nothing else, and every finding must quote the passage it rests on.
  • Verifier agent gets the record and the reviewer’s claims, never its reasoning, and must keep, weaken or drop each one with its own evidence.
  • Human sign-off: a QA reviewer accepts, rejects or edits every finding and must write a reason. A hook on the closing agent refuses to write the record unless every reason is present and appears word for word.
How the Deviation Review Agent works: a reviewer agent reads one record, a verifier agent tests its claims, a human signs off every finding with a reason, and a hook refuses to close the record if any reason is missing. Deviation record 1 file, empty folder Reviewer agent reads one file only quotes every finding Verifier agent sees claims, not reasoning: keep, weaken or drop Human sign-off accept, reject, edit a reason for every finding Gateon close Record closed DENIED: a reason missing, no file written ALLOWED only when every finding is signed and every human reason appears word for word
agenthumangateThe record cannot close until the human step is complete.

93-second narrated walkthrough. Synthetic records throughout.

What was measured

12 synthetic records for a fictional company, 26 planted flaws (15 serious), 3 clean records including one trap. Scored against a fixed answer key.

serious flaw caughtmissedfalse alarm
ConfigurationSerious flaws caught (of 15)All flaws (of 26)False alarmsClean records passed (of 3)
Reviewer only1522112
+ verifier, balanced kept1522102
+ verifier, strict131952

Single runs on a small synthetic test set, so indicative only. Matching was done by an AI judge whose decisions are on file. The false-alarm trap (a late closure with an approved extension) passed in every configuration. The clean record that drew false alarms was an environmental-monitoring excursion.

Sample finding with human sign-off: DEV-2026-008
DEV-2026-008 · finding 1RC_UNSUPPORTEDprimary
“The investigation concludes that the operator was distracted during the wet-massing step and did not stop the mixer at 8 min, then recorded the intended time rather than the actual time in the batch record.”

Operator distraction is asserted with no supporting evidence, such as an interview about what distracted the operator, an observation, or a PLC event log. […]

accepthuman sign-off

“No evidence for distraction shown. No proof that the button is working correctly”

Tamara Levine · 12 Oct 2026, 10:01 UTC
One real finding from the kept configuration, with the quoted passage it rests on and the signed reason, copied from the run files. The record is synthetic.
Close record DEV-2026-008
DENIED   record not written
         finding 4: reason missing or shorter than 10 characters

Close record DEV-2026-008, after sign-off
ALLOWED  record written
         4 of 4 findings signed; 4 human reasons found verbatim
EL-03 · workflow

Reduced-Testing Writer

RecordEL-03
KindWorkflow with AI steps
Data21 synthetic evidence packs
Halts onMissing evidence

Full monograph testing of every incoming raw-material batch costs real lab time. A justified reduced or ID-only scope for a qualified supplier saves it, but the justification means pulling specs, three batches of in-house and vendor results, investigation history and vendor status from several systems. So in many labs it simply waits.

What it does

Given one material’s evidence pack, it proposes FULL or REDUCED for every test, classifies the material as ID-only, partially reduced or full testing, and writes a report for a QA reviewer in which every sentence points to its source. If evidence is missing or contradictory, it writes no report at all, only a missing-evidence notice.

The design decision

AI reads what code cannot, proposes what a human must approve, and writes what code has already decided.

Every decision (matching, unit conversion, spec tightness, within-spec checks, eligibility, classification, halting) is tested Python on rules a reviewer can read and edit. AI is used in five places only: reading vendor certificates of analysis in different layouts, proposing name mappings for a human to approve, drafting a method-equivalence rationale for a human to decide, drafting the report text, and checking that text against the evidence. Identity and description are never reduced. A vendor or in-house result out of spec stops the whole abbreviation.

How the Reduced-Testing Writer works, in three lanes. AI reads CoAs, proposes name mappings, drafts an equivalence rationale, drafts the report and checks it. A human approves mappings, decides equivalence and reviews the report. Python rules make every decision, including halting with a missing-evidence notice. AI steps Python rules Human Evidence pack in spec, results, CoAs, history 1 Read vendor CoAs 2 Propose name mappings Approve mappings 3 Draft equivalence rationale Decide equivalence Decide every scope, classify or halt 4 Draft report text 5 Check text against evidence QA reviewer reads report Missing-evidence notice no report written halt
AI step (5)Python rulehumanhaltFixed order, wired by code. The AI never takes a decision.
27/27
Whole packs with AI in the loop: 9 packs, 3 repeats each, every run matched the answer key.
What was testedResultNote
Deterministic rules110 tests passNo AI in these
Whole packs, with AI in the loop27 of 27 runs9 packs, 3 repeats each, matched the answer key
Reading synthetic CoAs480 of 480 fields12 CoAs in 3 generated layouts (easy case)
Reading real public CoAs48 of 48 (larger model)
45 of 48 (smaller model)		5 public vendor CoAs, not republished here; labels by a QC expert
Claim checker on planted false claims71 of 72 caughtWrongly flagged 5 of 72 true claims
Missing or contradictory evidencehalt packsHalted correctly, before any AI step was called
Sample report, raw output: pack P01 (synthetic)

Reduced-testing justification: Sodium Citrate Dihydrate (invented)

Pack P01_approve_full | evaluation date 2026-10-15 | ruleset rs-0.1-draft | spec revision R4
1. Decision summary

Material classification is FULLY. M-class 2 tests are proposed FULL and 8 REDUCED, so the testing mode is ID_ONLY. M-counts, M-mode Next periodic review date is 2027-03-10. M-review

2. Per-test proposed scope
TestIn-houseVendor name as printedMappingScope
T01 DescriptionVISUALAppearancealiasFULL T01-scope
T02 Identification A (IR)IRIdentification (IR)proposed-approvedFULL T02-scope
T03 Assay (anhydrous basis)TITRATIONAssay (anhydrous basis)exactREDUCED T03-scope
T04 pH (5% w/v solution)POTENTIOMETRYpH, 5% solutionproposed-approvedREDUCED T04-scope
T05 Water (KF)KFWater by KFproposed-approvedREDUCED T05-scope
+ 5 more tests, all REDUCED; then the three-batch comparison
FULLY means fully reduced: every test except identity and description is reduced, so the testing mode is ID_ONLY. Run without the model: the decisions are the same Python, and the wording is the fixed template, not the AI draft. Pointers like T01-scope tie each line to its source.
Missing evidence notice (no report written)
SYNTHETIC DATA. PROTOTYPE ONLY. NOT VALIDATED.

Halt code: MISSING_COA
Detail:    no CoA for in-house batch NB-2603 (vendor batch AL-5503)

The writer halts rather than write around missing or
contradictory evidence. Supply the missing item and rerun.

Questions

Who builds the tools?

The rules, test sets, answer keys and every design decision come from hands-on experience in pharma QC and QA. The code is written with AI and tested against them, and every output is reviewed by a human.

Is any of this validated?

No. These are prototypes on synthetic data, not validated for GMP use. Any lab use would need validation under the lab’s own CSV process.

Can we try one?

Yes. Send a short description of the problem and the pilot is scoped together: which tool, which data, and what a useful result would look like.

Why not just use an AI chat like Copilot, ChatGPT or Gemini?

For drafting text, often you should. These tools are about the parts a chat assistant does not give you: fixed rules that a QA reviewer can audit, refusal when evidence is missing, and a record of what the human decided and why.

Contact

Have a lab problem like these, or want to try one of the tools on your own data? Send a few lines about it.

Based inKfar Saba, Israel