How V5 Ultimate runs an ingredient blender, end to end.

Ingredient and dry-mix manufacturers are the upstream node for almost every food, beverage, supplement and pet brand. A single undeclared peanut, a mycotoxin excursion in a corn-starch lot, a Salmonella positive in a spice blend, or a non-organic substitution in a certified-organic premix becomes the customer's recall — and the blender's reputation event.

21 CFR 117 (FSMA Preventive Controls) is the operating regulation. FSMA 204 applies to any Food Traceability List ingredient passed through (e.g. shell-egg powder, certain leafy-green powders, certain cheeses, certain nut butters). FSVP (21 CFR Subpart L) applies to imported ingredients. GFSI codes — FSSC 22000, SQF, BRCGS — sit on top, and customers audit against them quarterly.

V5 Ultimate treats every supplier lot, every dispense, every blend cycle, every allergen-changeover sanitation, every QC result and every CoA issued as part of the same immutable ledger. Recall is a query. Allergen claims and organic / non-GMO / kosher / halal certifications are bound to the recipe and the lot — they cannot drift.

Every node carries forward supplier lot, work order, operator, equipment, allergen profile and certification (organic / non-GMO / kosher / halal) context from the node before it. Nothing is re-keyed.

21 CFR 117 Subpart G (Supply-Chain Program) and 21 CFR Subpart L (FSVP) require the blender to control hazards before the ingredient arrives. V5 holds every supplier as a structured object: scope by ingredient × grade × site, allergen profile (Big 9), kosher / halal / organic / non-GMO certificate status, mycotoxin / heavy-metals / micro CoA history, FSVP hazard analysis and verification activity per imported ingredient.

Re-qualification is calendar-driven and portal-led; an expired allergen statement or FSVP file blocks goods-in automatically. Supplier risk scoring drives the depth of incoming inspection.

Ingredient blending is food, not pharma. 21 CFR 117 doesn't require you to quarantine every incoming lot until QA release — it requires receiving against written specs, approved-supplier criteria, and the controls your hazard analysis says are needed. V5 mirrors that. The receiver scans the ASN, weighs against PO, captures the supplier lot, traceability lot code (TLC for FTL items), CoA, allergen statement and best-before. V5 mints a goods-receipt lot ID and, unless a hold rule fires, the lot is immediately available to production.

Hold is configurable per ingredient and per supplier — examples: missing or out-of-spec CoA, supplier on probation, identity verification due on this lot, temperature excursion at receipt, mycotoxin- or heavy-metals-sensitive ingredient flagged for incoming test, or country-of-origin watchlist hit. Anything else flows.

Identity verification — visual, NIR/FTIR for botanical and chemical identity, ELISA for species in animal-derived ingredients, mycotoxin rapid test on corn / wheat / nuts, heavy-metals screen where required — runs on the same kiosk on the cadence your plan dictates. Results link to the receipt lot, instrument, analyst e-signature and supplier CoA.

For any FSMA 204 FTL ingredient, the Receiving event KDEs (TLC, source ref, date, location, quantity) are captured to the immutable ledger whether the lot is held or not.

V5 maintains bin attributes: allergen profile (Big 9), organic segregation, kosher / halal segregation, temperature band, humidity. The pick list shown to the picker only includes bins compatible with the work order — peanut-bearing ingredients cannot be picked into a 'peanut-free' production run; conventional bins cannot be picked into an organic blend. FEFO is enforced; overrides require reason + dual e-sig.

Each recipe holds: BOM with ingredient × supplier × allergen profile per line, addition order, blend routing (mixer type, speed, duration), in-process control plan (mesh, micro panel, bulk density, mycotoxin, heavy-metals), finished-product spec, CoA template, packaging routing, allergen statement, certification claims (organic / non-GMO / kosher / halal), and the locked label artwork version.

Effectivity is two-signature: a formulator signs the change, a PCQI signs the validation impact. New work orders pick up the new revision; in-flight WOs keep the snapshot they were released against.

If the recipe is revised mid-shift, the work order keeps the snapshot it was released against. The auditor sees exactly the formulation this batch was built to, with the sha256 of the snapshot recorded against release.

Dispense is scale-stream — the kiosk reads the weight, verifies within tolerance, captures the ingredient lot scan and operator e-signature. Out-of-tolerance trips an in-process deviation. The blend step shows the recipe's mixer speed, time and addition order; deviations are flagged in real time.

Allergen changeovers are explicit, signed gates. A peanut-bearing run followed by a 'peanut-free' run requires a documented changeover sanitation (wash-water analysis or ATP swab or visual + ATP) with the result bound to both work orders and a dual e-signature before the next run releases.

Bulk density, mesh / particle size, moisture, micro panel (Salmonella, E. coli, indicator organisms), mycotoxin on grain / nut / spice blends, heavy-metals where the spec commits, allergen ELISA on changeovers, vitamin / functional-actives assays for premix lines — all scheduled by the work order. OOS auto-opens a deviation, holds the lot, and notifies QA.

A single register categorises by hazard type. Root-cause uses the structured fields the auditor expects. CAPAs link to the deviation, affected lots, operators, equipment, complaints, and Reportable Food Registry where required.

Bag / tote / IBC labels carry product name, lot, manufacture / best-before dates, allergen statement, certification marks (organic, non-GMO, kosher, halal), country of origin and net weight. The CoA is generated automatically from the QC results bound to the work order — every spec line, every test, every analyst signature.

Where the product contains FSMA 204 FTL inputs, the Transformation event KDEs (input TLC(s), output TLC, date, location) are captured and the Shipping event propagates the new output TLC to the customer ASN.

Review-by-exception is only safe if the record is complete and immutable. V5's batch record is built from the ledger as the line runs. The QA reviewer's queue surfaces only items the system can't auto-clear. Release requires two qualified e-signatures, both identity-verified, both bound to the snapshot.

Every shipment goes out against an ASN that carries lot codes (and FSMA 204 TLCs where applicable). When a recall trigger fires — internal QA disposition, supplier withdrawal, customer complaint, FDA Class I — V5 walks the ledger in both directions in seconds: every batch made from this supplier lot, every customer DC it shipped to, every adjacent batch that shared the blender.

The customer-notification packet, the Reportable Food Registry submission (where applicable), the recall press release draft, and the mock-recall test for the next GFSI audit all assemble from the same ledger.

Hazards link to preventive controls, monitoring, corrective actions, verification, and the PCQI who owns each. Re-analysis is calendar-driven and prior monitoring data assemble into the re-analysis workspace automatically.

Draft → Review → Approved → Effective → Periodic Review → Superseded / Retired. Two qualified signatures effect every release.

V5 maintains a training matrix per role × line × hazard. PCQI training (21 CFR 117.4) is bounded; an expired course locks the operator out of the relevant kiosk action.

Every batch's quality signals, every CAPA, every supplier scorecard, every internal-audit finding, every customer complaint, every EMP swab — all stream into the management-review workspace as they happen.

V5's audit module schedules internal audits and external audits (FSSC 22000, SQF, BRCGS, customer 2nd-party). When FDA arrives, the inspector's request list — Food Safety Plan, last three months of monitoring, corrective-action log, supplier verification — is a query, assembled and printable in minutes.

Hazards are identified by category (biological, chemical including allergens, physical, radiological). Each gets a severity × likelihood score, a documented decision on whether a preventive control is required, the monitoring procedure and frequency, the corrective action, and verification.

Big-9 inventory of allergens used on site, allergen-zoned warehouse and production lines, allergen schedule (peanut-free runs on Mondays, peanut-bearing on Wednesdays), changeover sanitation procedures, ATP / ELISA verification, label-on-pack reconciliation at packaging — all assembled into one programme with a calendar and an audit trail.

V5 holds sanitation procedures per zone × equipment × allergen state, with frequency, agent, concentration, contact time and verification (visual + ATP + microbiological swab on a defined cadence). Sanitation events are kiosk-led; results bind to the equipment record.

Dispense scales, blend-cycle timers, sieve mesh sizes, metal detectors, sifter magnets, moisture meters, particle-size analysers — each is a tracked asset with calibration interval, certificate, tolerance, and an out-of-tolerance impact-assessment that triggers a back-look across every batch since the last passing calibration.

Sample sites, frequency, indicator and pathogen testing, and corrective actions on a positive (vector swabbing, intensified cleaning, re-test) are all scheduled and tracked.

Intake captures customer, lot, best-before, place of incident, complaint type, severity. V5 binds the complaint to the batch, surfaces adjacent batches that share the blender, and trips a deviation if pattern crosses a threshold.

Every signed event carries the actor's verified identity, signature meaning, time, and a sha256 of the record. The signature row is append-only. PDF renders are deterministic from the snapshot.

CoAs per lot, allergen statements, organic / non-GMO / kosher / halal certificates, FSVP attestations, country-of-origin documentation — all on the portal. Audit-trail of who downloaded what is automatic.

A typical plant onboards in 4–8 weeks: master data import, recipe migration with allergen-profile mapping, kiosk training, parallel-run validation, cutover. The validation pack (URS → IQ → OQ → PQ) is built in.

Abbreviated mapping. Each link opens the full readiness guide.