Skip to main content

Master recipes. Versions. Changes. Second lines. Recipe management. Every limit, with its evidence.

Each step of a master recipe keeps its limits, the study or procedure they rest on and the lots made to them. A change is a new version, tested before a lot runs it. Neil proposes; your team approves.

Master recipe MB-24 version 7 in Seal, mAb-24 for Rookery Bio on Line 2, on Tue 17 Nov: nine stages and 26 steps, each with its limits. Step 3.1 is open. The bulk may be held 36 hours before filtration; the study it rests on, EXP-332, covers holds of 6 to 26 hours. Of the 14 lots made to version 7, lot 26-0917 was held 29 hours 40 minutes, past the study and inside the limit, which Neil found on Mon 9 Nov and deviation DEV-041 assessed. Version 8, approved, takes effect from lot 26-0923. The fill weight check rests on procedure SOP-FIL-007, and each lot keeps the version of it the lot was made to; the hourly bowl check came in with its version 13, from lot 26-0918.
Recipes › MB-24
Effectivev7

MB-24 mAb-24 for Rookery Bio, Line 2

v8 approved, from lot 26-0923
Steps of v726 steps, 14 lots made to it

1 Dispense

  • 1.1Line clearanceSigned by one person, verified by another
  • 1.4Materials scannedEach container released, in date and reserved for the lot
  • 1.6Drug substance weighedTo the lot's calculated weight, within 0.05 kg
  • 1.7Dispense reconciledIssued, used and returned add up

2 Formulate

  • 2.1Vessel V-201 readyUsed within 72 h of its cleaning
  • 2.6pH and osmolalitypH 5.8 to 6.2, 280 to 320 mOsm/kg
  • 2.9Bulk completeFiltered within 36 h

3 Filter

  • 3.3Bioburden sampleUnder 10 CFU in 100 mL

4 Fill

  • 4.1Start of fillF-⁠02 on fill program FP-⁠24 v3; everyone who signs trained
  • 4.4InterventionsStart, end and vials of each, before the next entry
  • 4.5Environmental monitoringGrade A, each minute of each run
  • 4.6End of fillVials filled, rejected and sampled add up

5 Cap

  • 5.1Caps loadedA released cap lot, in date
  • 5.2Capping completeCrimp checked on 5 vials every 30 min

6 Inspect

  • 6.1Visual inspectionEvery vial, rejects counted by defect
  • 6.4Release samplesSamples for 19 release tests, labelled to the lot

7 Label

  • 7.1Labels issuedThe lot's text, checked against its approved proof
  • 7.3Labels reconciledIssued, used and destroyed add up

8 Pack

  • 8.1Cartons completeEach weighed on the checkweigher
  • 8.2Shippers completeCounted to the lot

9 QA review

  • 9.1Review by exceptionBlubber's 50 checks, each met or decided by QA

Step 3.1 Filtration

MB-24 v7
36 hStudied to 26 h

Bulk held at most, from step 2.9 to the start of filtration

Lots made to v7EXP-332, 6 to 26 h
Past it
Past 36 h the fill can't start without QA
Rests on
EXP-⁠332 Bulk hold: six holds of 6 to 26 h at 2 to 8 °CDEV-⁠041 Lot 26-⁠0917 held past the study, accepted by QA
Read from
The record's own times, step 2.9 to step 3.1
Each limit with what it rests on, and the lots made to itLot 26-⁠0917, Fri 6 Nov: held 29 h 40 min, 3 h 40 min past EXP-⁠332. Found by Neil Mon 9 Nov 22:15; DEV-⁠041Each lot keeps the version of SOP-⁠FIL-⁠007 it was made tov13's hourly bowl check, on every lot from 26-⁠0918Each limit with what it rests on, and the lots made to itEach limit with what it rests on, and the lots made to it

A change is a new version, tested before a lot runs it.

Neil proposed v8 from DEV-041's evidence. Its tests were approved before they ran, and every lot made to v7 keeps v7.

Changeset 17 in Seal: master recipe MB-24 from version 7 to 8, on Tue 17 Nov. Step 3.1 keeps its 36 hour limit, and a bulk held past 26 hours now waits for QA before filtration, because study EXP-332 covers 6 to 26 hours; deviation DEV-041 found the gap. Neil proposed it on Fri 13 Nov. R. Haldane, Validation, approved its three new test cases before they ran, and 16 of 16 test cases passed. K. Brandt, Production, approved on Mon 16 Nov and S. Amato, QA, at 11:05 today. Version 8 takes effect from lot 26-0923; lots 26-0909 to 26-0922 keep version 7.
Recipes › MB-24 › Changeset #17
Approvedv8

Changeset #17 MB-24 v7 to v8

Proposed by Neil from DEV-041
  1. Proposed
  2. Tested
  3. Signed
  4. Effective
Step 3.1 Filtrationv7 to v8
Limit
Bulk held 36 h at most
Past 26 h
Filtration waits for QA's approval
Past 36 h
The fill can't start without QA
Why
EXP-332 covers holds of 6 to 26 h; its extension to 36 h is under way. DEV-041 found lot 26-0917 held 29 h 40 min.
Each lot's versionv8 from 26-0923

Effective for lots created from Wed 18 Nov, the first 26-0923. Lot 26-0922, filling today, keeps v7.

Before v8 takes effect

4 of 4
Proposed by Neil, Fri 13 Nov 15:40, from DEV-⁠041Test cases approved first, then run on v8Each signature a name, a time and a meaningEvery lot keeps the version it was made to

F-02 runs the fill. Seal only reads it.

MB-24 names F-02's fill program and its version. Seal Edge reads what F-02 loaded and did, and sends it nothing.

Lot 26-0918 on filler F-02, the night of Wed 11 Nov. Master recipe MB-24 version 7 authorised fill program FP-24 version 3, 75 vials a minute and procedure SOP-FIL-007 version 13. Seal Edge read F-02: the program loaded at 00:17 matched, the fill started at 00:20, the link to Seal was down from 01:12 to 01:26 and 33,600 readings were held and sent, and pump 2 was trimmed at F-02's panel under D. Osei's login at 02:14. Seal sent F-02 nothing.
Lots › 26-0918 › Fill
MB-24 v7

Lot 26-0918 on F-02 Wed 11 Nov

Read-only
Authorised by MB-24 v7, step 4.1
FP-24 v3F-02's fill program. F-02's controller runs it
Line speed
75 vials a minute
Fill weight
10.20 g, each of 4 pumps checked every 30 min
Procedure
SOP-FIL-007 v13, in force from Tue 10 Nov 21:34
Read by
EDGE-L2, 31 of F-02's tags over OPC UA

Read from F-02

Seal Edge

Sent to F-02 by SealNothing

The program F-⁠02 loaded, against the one MB-⁠24 namesF-⁠02 runs the fill; Seal Edge reads it each secondA dropped link loses nothingYour people run the line from F-⁠02's own controls

A second line starts as a copy and its differences.

Line 1's copy of MB-24 keeps the steps, limits and studies, and none of the sign-offs. Neil lists what differs, each with an owner.

MB-24 for Line 1 in Seal, a draft copied from Line 2's version 7 by K. Brandt on Mon 16 Nov. It keeps the steps, limits, studies and procedures, and none of the sign-offs. Neil lists five differences from Line 2, each with a proposed owner: F-01's two pumps against step 4.2's sampling of four; F-01's slower fill and no study of the filtered bulk held that long; balance B-03's calibration not linked; CAPA-016 open on washer W-01; and EDGE-L1's tags on F-01 not mapped to the fill's steps. The copy approves no equipment, validation or readiness.
Recipes › MB-24 for Line 1
Draft

MB-24 for Line 1 from Line 2's v7

Copied by K. Brandt, Mon 16 Nov
Carried from Line 2's v7
  • 9 stages and 26 steps, with their limits
  • Studies EXP-⁠318, EXP-⁠321 and EXP-⁠332
  • SOP-⁠FIL-⁠007 v13 and WI-⁠FIL-⁠011 v3
Left with Line 2's v7
  • Its approvals
  • Its test results
  • Lots 26-⁠0909 to 26-⁠0922, made to it
Bound toLine 1Line 2
FillerF-⁠01, 2 pumpsF-⁠02, 4 pumps
IsolatorISO-⁠01ISO-⁠02
BalanceB-⁠03B-⁠04
Seal EdgeEDGE-⁠L1EDGE-⁠L2

Differences Neil found

Proposed owner
  • F-⁠01 has 2 pumps. Step 4.2 samples 5 vials from each of F-⁠02's 4K. Brandt
  • F-⁠01 fills 50 vials a minute, not 75, a fill half as long again. No study MB-⁠24 cites covers itN. Farrow
  • B-⁠03, not B-⁠04, weighs the fill check. Its calibration isn't linked yetT. Nakamura
  • Line 1's washer W-⁠01 has CAPA-⁠016 open: needle blockage, 6 of 10 lotsT. Nakamura
  • EDGE-⁠L1 reads F-⁠01; its tags aren't mapped to steps 4.1 to 4.6C. Whitlock

The gap assessment goes to each owner. Nothing in the copy is approved.

A copy of v7: its steps, limits and referencesNeil compares the copy with Line 1's recordsEach difference points at both linesEach difference points at both linesEach difference points at both linesA copy approves no equipment, validation or readiness. The owners decide

Questions about recipe management.

What is a recipe in Seal?

The master batch record is executable configuration in Seal: instructions, required fields, calculations and review steps on the same platform as materials, equipment, QC and quality. Seal represents it as reusable stages and steps with explicit behaviour rather than page layout.

What does a step check, and what happens past a limit?

The approved workflow checks each prerequisite before a step starts: procedure version, operator qualifications, material and equipment status, and line clearance. Each controlled value has a range and a configured response: continue, require verification, repeat the observation, create a sample, place the batch on hold or open a deviation. The original value and the system’s evaluation stay visible even when the authorised process continues.

Can a limit keep the study behind it?

Yes. A critical process parameter keeps its justified range, the quality attribute it controls and the studies that established it. Development and manufacturing share linked definitions and evidence in Seal, so a manufacturing recipe can reference the runs behind each parameter. Development results do not by themselves authorise manufacturing execution.

What happens to lots made or in progress when the recipe changes?

Seal resolves effectivity for the product, site, market and start date when the batch is created, and records the versions selected. A later change does not rewrite work in progress. If an urgent change must apply to live work, authorised users define the transition point and approve the additional instructions, and both the original and changed course remain visible. A revised route, limit or yield rule becomes a new recipe version for the lots that follow; every earlier lot keeps the recipe it was made under.

How is a change to a recipe tested?

A finding becomes a new version of the process, carrying its requirements, tests and approval. Selected user acceptance tests (UATs) run in a background simulation when a change is prepared for review. Publication requires current passing results for the required UATs; missing or stale results are not started automatically. Calculations such as yield, potency adjustment or material balance are versioned process logic, so a changed formula is a controlled revision tested against representative cases before it takes effect.

Can Neil write the recipe from our master batch record?

Give Neil a procedure or master batch record. It proposes the steps, fields, calculations, material and equipment requirements, and review gates in Seal. Process experts check the source mapping and resolve assumptions before testing and approval. Nothing Neil produces takes effect until a qualified person has reviewed it, and operators run only the approved version.

Does Seal download recipes or send setpoints to our equipment?

No. Every connection Seal Edge makes is read-only, and nothing in Seal can send a value, a setpoint or a command to a machine. Historian and control-system connections read values; they do not change setpoints, acknowledge alarms or operate equipment. DCS, PLC and historian systems remain authoritative for process control.

How does a recipe move to another line or site?

A permitted copy of the configured process starts the receiving site as a separate draft. It keeps the steps, fields, calculations and references, and none of the sign-offs. A site-specific process binds the platform process to that site’s equipment, scale and facility constraints, so the differences from the source are recorded rather than implied. Copying is not automatic site readiness.

Can we start without replacing the MES we run?

You can start with one process, alongside the MES, LIMS or QMS you already run. Once we agree the process evaluation scope and receive the required inputs, see the first working configuration and supporting validation pack within 48 hours.

All 11 capabilities
Master recipes as executable steps
The master batch record is executable configuration in Seal: instructions, required fields, calculations and review steps on the same platform as materials, equipment, QC and quality. Seal represents it as reusable stages and steps with explicit behaviour rather than page layout.
Instructions from the definition
The displayed instruction is generated from the structured requirements, so the readable record is derived from the same definition as the executable logic rather than maintained separately.
Limits with configured responses
Each controlled value has a range and a configured response: continue, require verification, repeat the observation, create a sample, place the batch on hold or open a deviation. The approved workflow checks each prerequisite before a step starts: procedure version, operator qualifications, material and equipment status, and line clearance.
Parameter evidence
A critical process parameter keeps its justified range, the quality attribute it controls and the studies that established it. Development results do not by themselves authorise manufacturing execution.
Controlled versions
Authoring is itself controlled: draft, review, approval, effective and superseded states have defined permissions, and comparisons, signatures and validation evidence stay with the version.
Changes tested before use
A finding becomes a new version of the process, carrying its requirements, tests and approval. Selected acceptance tests run when a change is prepared for review.
Effectivity and lots in progress
Seal resolves effectivity for the product, site, market and start date when the batch is created, and records the versions selected. A revised route, limit or yield rule becomes a new recipe version for the lots that follow; every earlier lot keeps the recipe it was made under.
Recipes proposed from your batch record
Give Neil a procedure or master batch record. It proposes the steps, fields, calculations, material and equipment requirements, and review gates in Seal. Process experts check the source mapping and resolve assumptions before testing and approval.
Line and site versions
A site-specific process binds the platform process to that site’s equipment, scale and facility constraints, so the differences from the source are recorded rather than implied.
Change impact
A new process or specification version does not silently change a batch already in progress; change control identifies the dependent records, training, inventory, open batches, labels and markets before the new state takes effect. Propagation creates tracked implementation work; it never silently edits an executing record.
Equipment stays in control
Historian and control-system connections read values; they do not change setpoints, acknowledge alarms or operate equipment. Readings from balances, scanners and equipment interfaces retain their physical source, while DCS, PLC and historian systems remain authoritative for process control and supply the events, critical values or alarms the record needs.

See your process in Seal.

Working 48 hours after we agree its scope: tested, with evidence your QA approves.