Process intelligence that stands up in an audit.

Limits with provenance, not limits in a spreadsheet

Every critical process parameter and critical quality attribute has a limit that lives in a controlled document. In-Sight extracts those limits from the batch record, the validation report, or the equipment datasheet, proposes them to a qualified person with the page cited, and arms them only when that person confirms. The limit, its source page, who confirmed it, and when, are all recorded. Nothing is auto-armed.

Alongside the documented limits, statistical baselines learned from the process itself catch the drift that stays inside specification but signals a change: a slower granulation endpoint, a lyophiliser that reaches shelf temperature later each cycle, a clean room differential pressure that recovers more slowly after each door event.

• Critical process parameters: temperature, pressure, pH, dissolved oxygen, agitation, feed rates
• Environmental monitoring: differential pressure, air changes, temperature and humidity by room
• Utilities: purified water and WFI loops, clean steam, HVAC performance
• Batch context: parameters tracked per batch and phase, with the batch as the investigation unit

Deviation investigation with an evidence chain

When a parameter leaves its limit or a baseline flags a drift, In-Sight opens a finding with the timeline already built: the parameter trend, the related signals, the operator round readings, and the alarms the control system raised. An investigation case rolls in related findings, ranks hypotheses by evidence, and lets the copilot draft the investigation narrative with citations to the SOP or batch record page it is drawing on. Where it cannot source a claim, it abstains.

The disposition writes back to every rolled-up finding and the case is tracked for recurrence after closure. The result is an investigation record with a complete chain from the raw data to the decision, and the decision-maker on it.

Continued process verification, without the monthly scramble

Continued process verification depends on trending every CPP and CQA over time and acting on the signal. In-Sight computes the SPC statistics continuously, generates the CPV report as a Word document on the schedule you set, and includes the findings raised and the investigations closed in the period. The report is the same one the team has been looking at all month, not a separate exercise.

Access control and audit trail

Every action in In-Sight is attributed: who confirmed the limit, who launched the investigation, who recorded the disposition, and when. Access is role-based at the workspace level and the platform can be deployed inside the site network or in a dedicated cloud environment. AI features can be pointed at an approved endpoint. In-Sight is designed to support a validated environment; the validation itself is done with the site's quality team.

Questions we are asked

Can In-Sight be used in a validated environment?

It is designed to support one: limits with provenance, attributed actions, a complete audit trail, and deployment inside the site network. Validation is performed with the site's quality team as part of the implementation.

Where do the limits come from?

From your controlled documents. Batch records, validation reports, and equipment datasheets are parsed, the limits are mined and proposed with the source page cited, and a qualified person confirms each one before it is armed.

Does the AI copilot make decisions?

No. It builds the evidence, drafts the narrative with citations, and ranks hypotheses. Every decision is made and recorded by a person, and the copilot abstains rather than guess when it cannot source a fact.

Which systems does it connect to?

OPC UA, OSIsoft PI, Siemens S7, Modbus TCP, EtherNet/IP, MQTT, BACnet for building and HVAC systems, SAP OData, REST APIs, and CSV files, with historical backfill.

Related