WorkflowAPI & Drug Substance

Process Development Report Authoring

A process development report in which each proposed range is drawn against the notebook, kilo-lab and plant runs behind it, each sentence opens the runs that prove it, and the client answers section by section in fields.

See one crystallization stage weighed against its runs ↓
Watch with transcript →
The problem

The edge a flask reached, written up as a proven range

A drug substance development report is assembled by one chemist from hundreds of notebook entries, the kilo-lab reports, the first plant batch records and a design study. Tables are retyped from spreadsheets. Proposed ranges creep wider than the runs behind them: an edge reached only in a one-liter flask is called proven, a strong word rests on a single run, and a sentence carried over from an earlier draft contradicts an experiment done since.

Each of those becomes a comment from the client's CMC reviewer, a question from the agency after filing, or a PPQ acceptance criterion the plant cannot meet. The same ranges are then typed again into the impurity risk assessment and the PPQ protocol, and when one document changes the others quietly fall behind.

Built forProcess development chemistsProcess development group leadsMSAT leadsAnalytical development scientists

The Problem Today

A range edge reached only in a flask gets written up as proven, and the client's reviewer finds it before the author does.

Each
studied range discussed against operating conditions (ICH Q11) published

Assembled by hand from

  • Notebooks
  • Kilo runs
  • Batch data
  • Old drafts
  • Lab edges sold as provenAn edge no kilo or plant run reached reads as a proven range
  • Three documents drift apartReport, impurity assessment and PPQ protocol each keep their own ranges

The Solution

  1. Each run placed on its rangeNotebook, kilo-lab and plant runs drawn at their values on each proposed range
  2. A disproved sentence opens its proofThe chromatogram that contradicts the draft sits beside the sentence
  3. An edge with no scale-up runNarrow it, call it lab scale only, or name the experiment still owed
  4. Released, then answered in fieldsThe group lead releases each section after reading it; the client accepts or questions

Built-in controls

  • Each sentence opens the runs behind it
  • The chemist decides each range; agents only propose
  • No one approves their own report

What You Get

One sourced set of claims carries the report, the impurity assessment and the PPQ protocol, and the next client's report starts from it.

  • A new chemist, a veteran's rulesThe plant's scale rule and past rulings apply to each draft
  • Claims the client can checkThe client's reviewer sees each claim with its runs attached
  • Weak edges found firstFound before the client's first review round or the agency
Features

A process development report with each range edge weighed against its runs

The salt formation and crystallization stage of one drug substance report, read by the agents, decided by the chemist, released by the group lead and answered by the client.

demoThe CDMO, its client, the people, the records and the figures are sample data.

  1. 01Range ladders

    Proven acceptable range and normal range drawn against every run by scale

    • Three lanes per parameter. Lab, kilo-lab and plant runs sit on separate lanes, so the chemist sees at once which scale reached each edge.
    • Hollow dots for failed attributes. A run that missed any quality attribute shows hollow and supports no edge.
    • The scale rule marked on each edge. An edge with no run at kilo or plant scale at or beyond it carries a gap mark under the MSAT lead's standing rule.
  2. 02Contradiction

    Contradicted sentences opened against the chromatogram that disproves them

    • Each sentence read against the runs. A claim of insensitivity that one notebook run and the design study disprove turns red before any reviewer reads it.
    • The proof opened beside it. The injection behind the finding opens in place, so the analytical scientist verifies the number without a search.
    • A sourced rewrite to accept or write over. The drafter proposes the sentence the data support, with its experiments cited; the chemist takes it or writes their own.
  3. 03Edge gap

    Range edges no kilo or plant run reached, flagged in the report text

    • Lab scale only, on the sentence itself. The drying sentence shows the gap beside the kilo runs it cites, so the chemist sees which words overreach before choosing a fix.
    • Weaker than its words. A sentence claiming uniform drying through the cake is held against the bottom-wall samples that say otherwise.
  4. 04Steer

    Steer the read in plain words and see which ranges lose their support

    • Old route runs left out on request. Scouting work from an abandoned route stops propping up the process the client will file.
    • Consequences shown on the ladders. The edges that lost their only support turn to no support at once, and the run log records who steered and why.
  5. 05Challenger

    A challenger that reads the draft as the client's CMC reviewer and the agency

    • Questions before the client asks them. Points the client's CMC reviewer or the agency would raise arrive with drafted answers while the section is open.
    • The agents' own links challenged. A kilo run seeded with the wrong lot is struck from the evidence, so the links the group lead relies on hold up.
    • Scale questions go to MSAT. A wash proven in a shallow kilo cake and assumed for a cake four times deeper goes to the MSAT lead to decide.
  6. 06Design study

    Design of experiments read in plain words, interactions included

    • The model fitted from the runs. Effects and interactions come from the recorded results, and the contour shows where the specification fails.
    • Interactions said in the report. A proven range for one parameter that holds only inside another's normal range is written that way.
  7. 07Decisions

    Every open range decision taken by the author, narrower or lab scale only

    • One queue of open items. Gaps, contradictions and weak sentences sit in one list, and submit stays blocked until each has a decision.
    • Each choice logged with its reason. The run that now supports each narrowed edge goes into the version history and the audit trail.
    • No range drawn wider. A decision can narrow an edge or label it lab scale only, so the report claims what the plant can run.
  8. 08Claim set

    One claim set for the report, the impurity assessment and the PPQ protocol

    • Three documents compared row by row. Where the PPQ protocol draft still carries an edge the report has narrowed, the row is marked before validation inherits it.
    • A change notice to the owner. The author sends the protocol's owner the one difference, and the protocol changes only when that owner edits it.
  9. 09Release

    Section release for client review, signed by the group lead after reading it

    • Read before release. The section must be opened and read by the group lead in person before the release dialog will sign.
    • Signed with a meaning. Name, time, reason and the signature's meaning stay with the record for the next audit.

The chemist decides the ranges; MSAT owns the scale claims

Who writes, checks and releases one development report.

Process development chemists

Write the report and decide every proposed range

  • Each edge drawn against its runs
  • Sourced rewrites for disproved sentences
  • Tables built from the notebooks
Process development group leads

Approve each report and release its sections to the client

  • Decisions to review, not assembly
  • A challenger that checks the agents too
  • A queue of reports with their clocks
MSAT leads

Agree the scale claims for the receiving plant

  • Their scale rule applied to each report
  • Cake depth and equipment fit questions flagged
  • Ranges the plant can actually run
Analytical development scientists

Verify the results the report quotes

  • The injection behind each number
  • A signature on results verified
  • Released results kept apart from in-process ones

A kilo run counted as support had been seeded with the old route's lot

Insensitive, said the draft

A notebook run at 45 °C with fast cooling gave Impurity B at 0.21% against NMT 0.15%; the design study showed the effect only when seeding low and cooling fast.

The wrong seed lot

A kilo run supporting the fastest cooling edge used the unmilled seed from the earlier route; struck out, that edge now rests on lab runs only.

A hold six hours long with data to four

The sulfonate ester rose about 0.5 ppm an hour during the hold, and no run at any scale went past 4 hours; the hold was narrowed to 2 hours.

A plant yield taken from the kilo lab

The draft gave 88–92% at plant scale; the five plant batch records read 86.4% to 89.7%.

demoSample data, from the working solution.

More screens

A bulk sign-off on routine sections waits until two drawn sentences are opened

Around the report: the client's portal, sign-off, the claims list, versions, the audit log and the scale rule.

A portal limited to this client's sectionsThe client's CMC reviewer answers each released section in fields; a free-text comment waits for the author or group lead and blocks approval until read.
Each signer signs only their own rowAnalytical development, MSAT, plant QA and the group lead each sign their own row; routine stages are signed by sample after two drawn sentences are opened.
Each sentence filtered by strengthThe 54 linked claims by stage and by strength, each with the runs behind it; a typed edit stays marked until a person reads it.
The stage paragraph, before and afterFour versions with who changed what and why; the old sentences struck through beside the new ones, nine of them changed this round.
Release, accept and range decisions in one logEach decision, release and answer with who, when, why and the signature meaning; exportable as a spreadsheet or an evidence binder.
Seed-grade rule held for the group leadThe scale threshold and rules learned from past reviews; a seed-grade rule proposed from the challenger's finding applies only once the group lead approves it.

demoSample data, from the working solution.

The report drafter proposes three fixes per gap and picks none of them

Seven agents read and propose; the chemist decides and the group lead approves.

Notebook reader

Reads each notebook experiment's charges, conditions, observations and results, and places it on each range it informs.

  • Changes no notebook
  • Drops no experiment because it failed
Batch reader

Reads kilo-lab reports, plant batch records and certificates, and checks seed lot, equipment and scale for each.

  • Cites no in-process result as a released one
Design-of-experiments analyst

Fits the design study and explains its effects and interactions in plain words.

  • Reports no model term the runs cannot show
Claim linker

Ties each sentence to the runs that prove it and marks claims that are weaker than their words or lab scale only.

  • Marks no edge proven without a run at kilo or plant scale
Contradiction finder

Reads each sentence against the runs and the model, and opens the record that disproves it.

  • Resolves no contradiction itself
Report drafter

Proposes a narrower range, lab scale only wording or a named experiment for each gap, and assembles the client package.

  • Picks no range
  • Sends nothing outside the company
Reviewer's eye

Plays the client's CMC reviewer and the agency against the draft and against the agents' own links.

  • Decides nothing
  • Lets no point drop unanswered

Read from the notebook, the chromatography system and the batch records

Electronic lab notebookEach experiment's charges, conditions, observations and in-process and final results.
Chromatography data systemThe injections and integration behind each result the report quotes.
Kilo-lab reportsScale, seed lot, filter and cake depth for each kilo run.
Batch records and released certificatesFrom the eBR, MES and LIMS: the first plant batches, development and clinical, with released results only.
The client's specificationThe limits each run is held against, owned by the client.
The documents that reuse the claimsThe impurity risk assessment and the PPQ protocol draft, read for where they disagree.

Process development report questions from chemists and MSAT leads

What is a process development report?

A process development report records how a drug substance process was developed: the route and its history, the critical quality attributes, each stage's studies and the proposed normal and proven ranges. For a CDMO it is owed to the client, and it underlies the manufacturing process development section of the client's filing.

What goes in a process development report template for an API?

Following ICH Q11: a narrative summary of attributes and parameters, the process history linking each batch to its stage of development, and each study with its ranges discussed against the proposed operating conditions.

When does a proven acceptable range count as proven?

Under the scale rule the MSAT lead set, only when a run at kilo or plant scale reached the edge and met every attribute. An edge reached only in the lab is narrowed, labeled lab scale only, or given a named experiment, and the chemist decides which.

How does a design of experiments change the report's wording?

The analyst fits the study and states interactions in plain words. When an impurity rises only with low seeding and fast cooling, the proven range for one is written with the other inside its normal range.

Who decides the ranges and who releases the report?

The process development chemist decides each range; the agents only propose. Analytical development verifies results, the MSAT lead agrees scale claims and plant QA checks the GMP batch data. The group lead releases each section and approves the report, and nobody approves their own.

How does the client review the report?

In a portal showing only that client's released sections. The client's CMC reviewer accepts or questions each one with a reason from a fixed list; a free-text comment goes to the author and blocks approval until a person reads it.

Is anything written back to the notebook or the plant systems?

No. They read the notebook, the chromatography system, LIMS, the batch records and the documents; nothing is written back to any of them. A change after a signature clears that section's signatures.

Bring one stage's
notebook runs.

Send one stage's notebook entries, kilo-lab reports and the draft section; we will draw each proposed range against its runs.