Procurement, decided in seconds.
Procurement teams across government and enterprise still source products the same way they did twenty years ago — manually cross-referencing sanctions registers, commodity exchanges, FX feeds, shipping schedules, supplier directories, and carbon registries to make a single defensible decision.
WYRM Sentinel replaces that workflow with a single agentic query. Proprietary AI agents work in parallel across every relevant data source, returning a ranked, scored, and audit-ready procurement recommendation in seconds.
The cost contrast
Comparable enterprise procurement intelligence stacks — combining spend analytics, sanctions screening, commodity data terminals, FX feeds, and supplier verification — typically run between £12,500 and £42,000 per month, and still leave the buyer to manually fuse the outputs.
WYRM Sentinel starts at £199 per month, with Pro at £499 per month — over 95% cheaper than the legacy enterprise stack — and delivers every input as a single agentic decision, with carbon scoring, situational analysis, and direct AI-assistant integration that standalone legacy tools do not offer at any price tier.
Our Story
Why we built this
WYRM Sentinel was built around a real-time multi-agent data platform — 50+ API routes, 20+ live data feeds, 10 specialised AI agents, and a geospatial analytics stack designed to surface material signals from across global open data sources.
That infrastructure proved to be the right architecture for the most acute and most underserved use case for real-time, cross-domain analysis: procurement and supply chain intelligence.
Today, WYRM Sentinel powers Supply-Root: an agentic AI procurement and supply chain intelligence product purpose-built for government and enterprise buyers operating under CBAM, sanctions, and supply-chain compliance regimes. Same engine. Sharper application.
Under the Surface
A specialist bench, not a search bar
Specialist agents (minimum 7 run per query)
Predictive models in the ensemble
Each agent reasons independently and adjusts its position as new data arrives. Their findings converge into a confidence score that adapts to market regime — and dissent between agents is surfaced rather than hidden, so every recommendation carries its own audit trail.
Founding Team
Leadership
James Reed
Founder & CTO
Full-stack engineer and Python specialist with a background in data engineering, automation, AI fintech systems, and multi-agent orchestration. Designed and built the WYRM Sentinel platform — the agentic AI engine that powers Supply-Root.
WYRM Sentinel operates as a standalone procurement and supply chain intelligence platform. Customer data stays within the Sentinel environment and is not shared with any adjacent infrastructure.
Technology
Built for procurement-grade reliability
Every architectural choice is made for buyers who need defensible decisions, not best-effort answers.
Live Procurement Feeds
Continuous ingestion across sanctions registers, commodity exchanges, FX markets, trade statistics, shipping lanes, and supplier registries. Two-layer deduplication for data integrity.
Parallel Agentic Workers
Seven specialised AI agents run simultaneously on every query — sanctions, compliance, commodities, FX, shipping, carbon, supplier verification — fused through a predictive scoring ensemble.
Audit-Ready by Default
Every decision logs the underlying signals, agent confidence scores, and ensemble weights. Buyers can defend any recommendation under procurement review.
Sub-Second Hot Path
Async Python on FastAPI with a Redis hot cache fronting Postgres. Repeat queries return instantly. New queries return in seconds, not minutes.
Multi-Channel Delivery
Web dashboard, REST API, and Model Context Protocol server. Buyers can search via UI, integrate via API, or query directly from Claude, ChatGPT, Grok, or Gemini.
Predictive Scoring Models
Custom AI models tuned on backtested procurement decisions. Confidence scores adjust to market regime. Dissenting signals are surfaced explicitly, never hidden.
Differentiation
What makes Supply-Root different
| Capability | Traditional Procurement Tools | Supply-Root |
|---|---|---|
| Decision speed | Hours of manual research | Seconds — single query |
| Sanctions checking | Manual portal lookups | Continuous, multi-jurisdiction |
| Carbon scoring | Spreadsheets after the fact | CBAM-ready at decision time |
| Supplier verification | Self-attested | Independently verified |
| Route + chokepoint risk | Not modelled | Live, agentic scoring |
| Situational analysis | Reactive only | Predictive simulation |
| AI assistant integration | Not available | MCP server included |
| Audit trail | Manual reconstruction | Auto-logged, defensible |