← Guides

How do you select AI manufacturing software for supplier quality oversight?

By Savi C., Co-Founder · Published Nov 17, 2026

What supplier quality oversight involves

Supplier oversight has two layers. The first is the supplier itself: selection, approval, quality agreements, audits and periodic requalification. The second is every lot that arrives: the certificate, the receiving record, any testing you do, and the release decision.

For cosmetics, ISO 22716 expects purchasing to be based on evaluation and selection of suppliers, and raw materials and packaging to be checked against acceptance criteria before release. It does not prescribe a specific identity test. Other sectors are stricter. Under 21 CFR 211.84, a drug manufacturer may accept a supplier report of analysis only if it runs at least one specific identity test and establishes the reliability of the supplier's results. Under 21 CFR 111.75, a dietary supplement manufacturer must test the identity of each dietary ingredient and may rely on certificates for other components only after qualifying the supplier and periodically reconfirming its results. Food facilities under 21 CFR 117 need a risk-based supply-chain program where a hazard is controlled by the supplier.

Most of the effort sits in the second layer. It is the same comparison, lot after lot, and it is where missing tests, changed limits and slow drift in supplier results get missed.

Which type of software covers which part

Supplier quality task

Approved supplier list and supplier status

Type of software that usually handles it

QMS or ERP

Supplier quality task

Quality agreements and audit records

Type of software that usually handles it

QMS or document management system

Supplier quality task

Supplier corrective actions (SCARs)

Type of software that usually handles it

QMS

Supplier quality task

Reading each incoming certificate and checking it against the specification

Type of software that usually handles it

AI review layer, or manual review

Supplier quality task

Comparing a certificate with the supplier's earlier lots

Type of software that usually handles it

AI review layer, or a spreadsheet kept by hand

Supplier quality task

Trending supplier results across lots and sites

Type of software that usually handles it

Analytics tools or an AI review layer

Supplier quality task

Lot release and supplier approval decisions

Type of software that usually handles it

Your quality unit. Software supports, people decide

Questions to ask in a demo

Bring a sample of real certificates, including scanned ones and a few you know had problems.

Reading the documents

  • Does it read scanned and handwritten documents as well as digital PDFs?
  • Does it handle certificates in each supplier's own layout, without a template per supplier?
  • Does it extract every reported test, method, limit and result, not just pass or fail?

Checking against your rules

  • Does it compare against your approved specification, at the current revision?
  • Does it flag tests that your specification requires but the certificate leaves out?
  • Does it flag supplier limits that are wider than yours?
  • Does it compare results with the supplier's earlier lots and show drift?

Control and evidence

  • Does every flag link to the page and line it came from?
  • Does release stay a human decision, with a record of who approved what?
  • Does it work with the QMS, ERP and paper records you already have?
  • Can it run where your data must stay, including on-premises if you need it?

How Litewave fits

Litewave is an AI review layer that works on top of your existing QMS, ERP, LIMS and paper records. Its Supplier Quality & CoA Review compares incoming certificates of analysis against your approved specifications and supplier history, and flags missing, changed or out-of-specification results, each linked to its source page. Quality Analytics trends results across batches and sites. It runs in Litewave's cloud, your cloud, or on-premises, with an air-gapped option. Litewave proposes, qualified people approve.

Our published results come from pharmaceutical manufacturing, and we make no claim about cosmetics, supplement or food results.

Questions we get asked

Do we need a new QMS to get AI for supplier quality?

No. Supplier approval, audits and SCARs can stay in your current QMS. An AI review layer reads the certificates and records you already receive and checks them against your specifications.

Can AI approve a supplier or release a lot?

It should not. Under ISO 22716 and the US drug, supplement and food rules, those decisions belong to your quality unit. Use AI to find problems and gather evidence for the decision.

Is supplier oversight different for cosmetics and drugs?

Yes. ISO 22716 asks for supplier evaluation and acceptance criteria but sets no specific identity test. 21 CFR 211.84 requires at least one specific identity test on each drug component even when you rely on the supplier certificate.

Sources

  1. ISO 22716:2007, Cosmetics, Good Manufacturing Practices
  2. 21 CFR 211.84, Testing and approval or rejection of components (eCFR)
  3. 21 CFR 111.75, Determining whether specifications are met (eCFR)
  4. 21 CFR 117.405, Requirement to establish a supply-chain program (eCFR)