Agentic AI
Agentic AI is an architectural pattern where autonomous specialist agents research, verify, and score decisions in parallel against live external data sources. Unlike retrieval-augmented generation, which retrieves from a static index, or copilots, which summarise the UI already in front of the user, agentic AI returns a ranked recommendation with a full evidence trail. Each agent owns a narrow domain; a fusion layer weights verdicts against the current market regime. In procurement, this means sanctions screening, CBAM carbon, FX, commodities, shipping, and supplier verification can all run concurrently on a single natural-language query.
Agentic Procurement
Agentic procurement applies agentic AI to buyer decisions. Instead of a procurement officer cross- referencing sanctions portals, commodity exchanges, FX charts, supplier directories, and CBAM calculators across a dozen tabs, agentic procurement runs specialist agents in parallel and fuses their outputs into one ranked, audit-ready recommendation per query. The evidence trail is the primary deliverable, which aligns with the UK Procurement Act 2023 expectation that supplier decisions be defensible on record.
CBAM (Carbon Border Adjustment Mechanism)
The European Union regulation that prices embedded carbon on imports of cement, iron and steel, aluminium, fertilisers, electricity, and hydrogen. From 2026 onwards, importers must surrender CBAM certificates matching the lifecycle emissions of in-scope goods. Procure computes CBAM liability at decision time using Climatiq lifecycle factors and country grid-carbon intensity, so buyers can compare origins on total landed cost including carbon.
OFSI (Office of Financial Sanctions Implementation)
The UK authority responsible for enforcing financial sanctions. OFSI maintains the UK consolidated list; inclusion prohibits UK persons from dealing with the designated entity without a specific licence. Procure screens every supplier and its declared parent ownership against the OFSI list on every decision, with revision diffs captured in the audit trail.
OFAC (Office of Foreign Assets Control)
The US Treasury authority that maintains the Specially Designated Nationals (SDN) list and operates US sanctions programmes. Non-US entities can still be caught by OFAC if they have US nexus — USD clearing, US customers, or US subsidiaries. Procure runs parallel OFSI and OFAC checks; conflicts and advisory-only findings are flagged explicitly rather than blocking silently.
OpenSanctions
A consolidated, open-data project that aggregates over 40 sanctions, politically-exposed-persons, and criminal-activity lists into a single queryable dataset, updated daily. Procure uses OpenSanctions as its primary sanctions-screening source, supplemented by direct OFSI and OFAC pulls for jurisdiction-specific metadata.
HS Code (Harmonised System)
The international customs classification standard, maintained by the World Customs Organization. HS-6 is the internationally common level; national variants extend to HS-8 (EU) or HS-10 (US, UK) with country- specific duty rates and preferential treatment. Correct HS resolution is the precondition for accurate landed-cost and CBAM calculations. Procure resolves HS codes automatically from natural- language procurement queries.
Total Landed Cost
The sum of unit price, freight, insurance, customs duty, CBAM liability, FX conversion cost, and any re-export or compliance fees. The only cost figure on which a procurement decision can be defensibly made. Procure returns landed cost in the buyer's reporting currency, broken down into each component, for every candidate origin and supplier.
Scope 3 Emissions
Indirect emissions in a company's value chain, including those embedded in purchased goods and services. For most organisations, Scope 3 is the largest share of the corporate carbon footprint. Increasingly subject to mandatory disclosure under CSRD, SECR, and SEC climate rules. Procure's carbon scoring surfaces Scope 3 exposure at the decision point so procurement can drive measurable reduction.
TM65 Embodied Carbon
CIBSE TM65 is the UK industry methodology for calculating the embodied carbon of mechanical, electrical, and plumbing (MEP) products when the manufacturer has not published an Environmental Product Declaration. It combines product weight, material composition, manufacturing-location grid intensity, transport, and end-of-life allowances into a single kgCO₂e figure per unit. Procure auto-generates TM65-compliant reports directly from a bill of materials, applying live country grid-carbon intensity from the same data layer that feeds the CBAM calculation. Output is an itemised workbook covering embodied carbon per product, total for the schedule, and a procurement recommendation for the lowest-carbon compliant option.
CPTPP
The Comprehensive and Progressive Agreement for Trans-Pacific Partnership — a free-trade agreement with preferential tariff concessions for originating goods. The UK acceded in 2024. Procure applies CPTPP preferences automatically when the origin, buyer, and HS code qualify.
AIS (Automatic Identification System)
The mandatory maritime tracking signal broadcast by commercial vessels over VHF. AIS data underpins shipping-lane analytics, chokepoint transit times, and dark-ship detection — vessels that deactivate AIS near sanctioned ports. Procure ingests AIS via AISStream, narrowed to the six active chokepoints (Hormuz, Red Sea, Malacca, English Channel, Panama, Gibraltar) for efficient coverage.
Chokepoint Risk
The sensitivity of a shipping route to disruption at a narrow waterway. Procure tracks the active status of the six global maritime chokepoints and scores every proposed route for exposure. Buyers can run what-if closures through Situational Analysis to quantify the impact of a blockage before it happens.
FX Forward Curve
The term structure of forward exchange rates, implied by interest- rate differentials and quoted in the market. For procurement commitments settling weeks or months forward, the forward rate — not the spot — is the economically correct reference. Procure uses ECB reference rates and forward curves to price FX exposure on multi- month commitments.
UN COMTRADE
The United Nations bilateral trade statistics database, covering over three billion records of reported goods flows between countries. The canonical source for origin validation, trade-flow benchmarking, and competitor-origin analysis. Procure uses COMTRADE to corroborate supplier-declared origins against reported trade flows — a mismatch is a flag.
OpenCorporates
The largest open database of company registry filings, covering over 200 million legal entities across 140+ jurisdictions. Procure uses OpenCorporates to verify supplier existence, trace ownership up the parent chain, and detect common-ownership across declared independent suppliers — a signal that often precedes sanctions-evasion structures.
Tier-N Supplier
A supplier N steps removed from the direct contractual counterparty. Tier-1 is your contracted supplier; Tier-2 is that supplier's supplier; Tier-3 is further upstream. Tier-2 and Tier-3 exposure is often where sanctions, CBAM, and modern-slavery risk materialise. A registered aluminium supplier shipping via a Turkish intermediary is routine. The same supplier invoicing through a new Turkish intermediary after a sanctions event targeting its parent is a signal that the confidence ensemble will weight accordingly.
SDN List
The Specially Designated Nationals and Blocked Persons List, maintained by OFAC. Inclusion means the designated party's assets are blocked under US jurisdiction and US persons are prohibited from dealings. The SDN list is updated continuously; Procure ingests every revision and recomputes affected decisions automatically.
Customs Code Resolution
The process of mapping a free-text procurement requirement (“15 kg raw aluminium”) to a jurisdiction-specific customs code (HS-8 7601.10.00 for the EU). Procure's natural-language layer resolves customs codes automatically, cross-checking against trade-statistics history to validate that the code is commonly used for the declared commodity.
MCP (Model Context Protocol)
An open specification for connecting AI assistants to external data sources and tools, originated by Anthropic and now supported across major AI clients. Procure exposes its procurement engine as an MCP server so buyers can query directly from Claude, ChatGPT, or any compliant client without leaving their existing workflow.
Audit Trail
The immutable record of every procurement decision, including the underlying data sources, agent outputs, confidence scores, ensemble weights, and the final ranking. Procure stores every decision in an append-only log; buyers can reproduce any decision and defend it under procurement review, internal audit, or regulatory inspection.
Confidence Ensemble
The mechanism by which Procure combines the outputs of multiple specialist agents into a single ranked decision with a confidence score. Each agent contributes a weighted signal; the ensemble adapts weights to market regime (stable, volatile, crisis). When agents disagree, the dissent is surfaced rather than hidden, with the supporting data points shown explicitly.
Situational Analysis
Scenario modelling against a live supply chain. Buyers feed in a hypothetical disruption — a chokepoint closure, sanctions expansion, commodity shock, CBAM tightening, or FX move — and Procure re-runs every basket and decision against the stressed conditions, quantifying landed-cost delta, CO₂ delta, and compliance-status change. Every projection is sourced and cited. See the agentic AI overview for the underlying architecture.
Feed Normalisation
The process of converting data from heterogeneous sources — OpenSanctions JSON, COMTRADE CSV, LME price series, ECB XML, AISStream WebSocket, OpenCorporates API — into a common schema so agents can reason across domains uniformly. Normalisation is a non-trivial prerequisite for cross-domain procurement decisions; without it, a sanctions flag cannot be automatically tied to a landed-cost recomputation or a CBAM delta.
Find a Tender
The UK successor to OJEU — the central portal for above-threshold public-sector procurement notices, run by the Cabinet Office. Ledger polls Find a Tender alongside Contracts Finder, SAM.gov, and BDUK feeds and scores each notice against the operator's capability profile and win history, returning a ranked shortlist with a fit score rather than a raw firehose of notices.
MTD / VAT (HMRC Making Tax Digital)
Making Tax Digital is the HMRC programme requiring digital VAT records and quarterly digital submissions. Ledger Enterprise produces MTD-compliant VAT submissions and reconciles customs entries against the same audit trail used for procurement and tender evidence — so the VAT, CBAM, and customs picture per transaction is one record, not three.
Tender Writer
Ledger's bid-drafting agent. Produces a structured first-pass response to a tender notice from the operator's evidence library and prior winning bids, aligned with the awarding authority's published evaluation criteria. Surfaces where claims are supported and where they require new evidence; output is a redlineable draft, not a finished bid.
CVE (Common Vulnerabilities and Exposures)
The public catalogue of known software vulnerabilities, maintained by MITRE and ingested by the NIST National Vulnerability Database (NVD). Each entry carries a CVSS severity score, affected products, and references. Cyber triages every new CVE against the operator's declared technology footprint and surfaces only the subset that actually requires action — usually a small fraction of the daily firehose.
EPSS (Exploit Prediction Scoring System)
FIRST.org's probability that a CVE will be exploited in the wild within the next 30 days. Where CVSS measures theoretical severity, EPSS measures real-world likelihood. Cyber uses EPSS to re-rank the CVE backlog — a low-CVSS vulnerability with high EPSS often outranks a high-CVSS one with negligible exploitation likelihood, because action is what matters.
KEV (Known Exploited Vulnerabilities)
CISA's catalogue of vulnerabilities being actively exploited in the wild. Where EPSS is predictive, KEV is observed. Cyber Enterprise raises KEV-listed exposures to same-day alert so operators can prioritise patching against confirmed in-the-wild activity rather than theoretical severity scores.
OSINT (Open-Source Intelligence)
The discipline of collecting and analysing publicly available information to answer an intelligence question. Cyber's OSINT toolkit covers username, phone, email, and image lookup across breach indices and OSINT sources, attack-surface mapping against operator-owned assets, and dark-web watch for credentials and brand mentions. Findings are surfaced as defensible evidence, not raw scrape.
MSA (Master Services Agreement)
The umbrella contract governing an ongoing commercial relationship, typically incorporating a schedule of services and statements of work (SOWs). Legal's clause-review agent produces clause-by-clause green / amber / red verdicts on MSAs against the operator's own clause playbook and surfaces the deviations that warrant a human redline.
DPA (Data Processing Agreement)
The contract module required under UK GDPR Article 28 (and EU GDPR equivalent) when a data controller engages a data processor. Legal flags DPA deviations from the operator's negotiated baseline and watches the ICO publication feed for guidance updates that may render existing DPAs stale — automatically tagging every affected contract in the library for re-review.
Clause Playbook
The operator's documented baseline positions on each contract clause — must-have, nice-to-have, and red-line. Legal compares every incoming MSA, NDA, framework, DPA, and SOW against the playbook and surfaces deviations with diff-style evidence and a suggested redline. Without a playbook, clause review is opinion; with one, it is policy enforcement.
Entity Resolution
The process of merging multiple records that refer to the same underlying real-world entity into a single canonical record. Data's entity-resolution agent unifies suppliers (from OpenCorporates), counterparties (from OpenSanctions), regulator-named entities, breach actors, and contract parties across every WYRM feed, with provenance per claim.
Cross-Domain Fusion
Routing signals between WYRM modules so a finding in one surfaces in the others. A sanctions hit in Procure triggers a Cyber supplier-attestation refresh and a Legal contract-review flag. A regulator update in Legal triggers a Procure compliance recheck. Data's orchestrator owns this routing and the shared audit trail so every cross-module action is reconstructable.