Understand the difference. Improve the process.

Investigations reuse process knowledge without prejudging cause. Event chronology, hypotheses, prior occurrences, process models, risk assessments, similar batches, equipment and material histories, analytical evidence, experiments, causal conclusions, affected population and actions remain traceable. Prior knowledge informs the investigation; it does not replace testing the current event.

Manufacturing support begins with complete context. Request, urgency, product, batch, operation, event, question, current state, production owner, quality involvement, technical owner, evidence request, actions, response, decision and closure define each support case. Tribal conversations can still happen; the technical decision and evidence cannot disappear with them.

The batch carries its readings, lab sample and investigation together. IPC-041 keeps its results, sampling context and source batch, and DEV-018 brings the batch and lab evidence into an open investigation; the deviation blueprint follows that investigation in full.

Batch context is assembled continuously. Materials, equipment, personnel, recipe execution, process signals, alarms, interventions, holds, samples, test results, deviations, environmental data, reconciliation and disposition remain connected before a review begins.

Keep the reviewed conclusion with the process version. A proposed change can carry its affected records and verification work forward through change control; it does not rewrite the original batch.

Manufacturing support brings batch execution, historian, laboratory and material evidence into the same technical investigation. A technology transfer compares the sending process with the receiving site’s scale, equipment, materials and methods. A trend review reads the trend with its process version, population and operating context, so the team investigates the signal, not just the chart. A proposed material, parameter or equipment change is traced to the processes and evidence it affects at each site.

Process ownership has a defined boundary. Product, process version, unit operations, site, line, scale, equipment train, materials, control strategy, validated state, markets, lifecycle stage, technical owner, quality owner, support model and effective dates define responsibility. The record distinguishes global process ownership from local execution ownership.

The process knowledge model carries intent. Material attributes, process parameters, in-process controls, quality attributes, ranges, normal profiles, interactions, models, failure modes, control responses, sampling, holds, equipment requirements and scientific rationale remain connected. Manufacturing can navigate from a limit to the evidence and decision that created it. ICH Q10 asks for exactly this: product and process knowledge managed from development through the commercial life of the product, drawing on development studies, technology transfer, process validation, manufacturing experience and change management.¹

Process models remain governed artefacts. Dataset, population, features, transformations, algorithm, assumptions, version, fit, diagnostics, prediction, uncertainty, intended use, applicability, validation, monitoring, reviewer and retirement define a model. A model that informs a manufacturing decision is not an untracked analyst file.

Commitments and residual risks remain visible. Validation commitments, enhanced monitoring, authority commitments, temporary controls, post-transfer actions, unresolved knowledge gaps, model limitations, due dates, owners, evidence, approval and closure stay attached to the effective process.

Process improvement has an evidence contract. Opportunity, hypothesis, expected benefit, risks, study or trial, protocol, batches, measures, acceptance, deviation, analysis, conclusion, change control, validation, filing, implementation, benefit verification and standardisation remain connected.

Process Development supplies the selected process, experimental evidence, scale assumptions and unresolved questions. MSAT assesses how that definition applies to the receiving equipment, site and operating conditions. A transfer package should make those differences inspectable; it does not establish equivalence just because the recipe names match. The purpose of transfer is to carry product and process knowledge between development and manufacturing, and between sites, as the basis for the manufacturing process, control strategy and validation approach.²

Technology transfer is an acceptance lifecycle. Sending definition, receiving fit, gaps, actions, document and master-data deployment, engineering batches, method transfer, training, qualification, validation, evidence review, commitments and acceptance remain one programme. MSAT can see which knowledge was accepted, adapted, rejected or remains provisional.

Site and line applicability is explicit. Facility, utility, equipment, automation, scale, geometry, sensor, recipe implementation, material source, analytical laboratory, staffing model, environmental state and local procedure define the deployed configuration. Site differences are assessed as controlled deltas rather than hidden inside local documents.

Site comparability separates equivalence from sameness. Process definition, equipment function, scale criteria, material sources, methods, normal profiles, capability, quality outcomes, deviations and uncertainty support the comparison. Different equipment can be comparable when the scientific and performance evidence supports the conclusion.

Manufacturing experience can then challenge the original understanding. Link a new observation to its unit operation, investigate it and prepare the proposed process or monitoring change. Preserve the earlier decision and evidence when a new version is approved.

Signals are evaluated against process knowledge. Alert, trend, drift, step change, atypical profile, capability decline, parameter-quality relationship, material shift, equipment behaviour, laboratory pattern or cross-site difference retains detection logic, population, baseline, magnitude, uncertainty, triage and consequence.

CPV is a technical decision system. Monitoring plan, parameters, attributes, stratification, frequencies, baselines, control limits, signal rules, exclusions, analyses, reviews, escalations, actions and reports remain governed. MSAT sees the signal with its process phase, material, equipment, site, method and batch context.³

When source evidence changes a parameter range, relationship, normal profile, control response or site applicability, the proposed updates appear as an explicit changeset. Reviewers accept, revise, reject or defer each change with rationale. The approved set publishes as a new knowledge version.

Changes traverse technical dependencies. Proposed change, rationale, affected products, sites, unit operations, parameters, attributes, materials, equipment, recipes, methods, validated state, stability, comparability, filings, inventory, supply, studies, actions and implementation state create the technical impact path.⁴

Approved knowledge propagates to controlled consumers. Approved updates identify affected recipes, master batch records, monitoring plans, validation requirements, specifications, training, methods, reports, transfer definitions, risk assessments and regulatory content. Propagation creates tracked implementation work; it never silently edits an executing record.

The operating loop is ingest, review, connect. Legacy files, source systems, new batch data, analyses, investigation conclusions and change evidence can propose structured updates. A human technical owner reviews the changes before approved knowledge becomes live. In line with the EU’s draft GMP Annex 22 on AI, Neil is not used in GMP execution. It helps set up configuration, which your team verifies and releases under change control. Customer data is not used to train AI models, and the requester’s permissions govern what Neil can read. See how Neil fits Annex 22 and more about Neil.

Seal acts as the cross-system technical record for products and processes. It does not replace MES, LIMS, QMS, historians, statistical tools or document systems; it connects their evidence to the current process model and the technical decisions made from it. A historian capture keeps its source timestamps, values, units and quality flags, so a reading in an assessment can be traced to the system that recorded it. See all integrations.

The first implementation should follow one product from transfer into commercial manufacture through a site-specific delta, engineering and PPQ batches, CPV signal, manufacturing support case, investigation, process-model update, change control, supplemental validation, controlled propagation and benefit verification.

Include conflicting source data, an unapproved model proposal, a parameter-range change, two sites with different equipment, an overdue commitment and a commercial batch made under the prior knowledge version. The system must preserve which knowledge was effective for every decision.