Procurement that runs itself.
Under your control.
supplyce is an agent fleet for source-to-pay. Named agents handle intake, sourcing, negotiation and invoicing inside guardrails you set — every action audited, priced and reversible.
EU-HOSTED · AUDIT-CHAINED · BYOAI
Every agent, one console.
Autonomy dials, kill switches, daily run caps, live traces and a cost ledger — the governed fleet at a glance, in the product you log into.
One loop, source to pay.
Eight areas, one governed flow — agents do the work, your policies decide what needs a human.
Describe it. It's a requisition.
Free text, an emailed quote or an uploaded PDF becomes a structured, policy-checked request — with per-field confidence you can see.
Let the agent haggle.
Round-by-round negotiation inside your target and walk-away limits — with a dry-run simulator and a full transcript.
Buy from catalogs you control.
Hosted catalogs with carts and preferred-supplier controls, plus real PunchOut round-trips over cXML and OCI.
Run competitive bids in minutes.
AI-drafted RFx with weighted scoring and award recommendations; live reverse auctions with anti-snipe protection.
Know every supplier. Watch every risk.
360° records, onboarding gates, maker-checker bank changes, continuous sanctions/PEP screening and 4th-party concentration maps.
Contracts that read themselves.
Clause extraction against your playbook, e-signature lifecycle, obligations that become tasks, renewals that never surprise you.
Touchless from PO to payment.
Three-way match with tolerances, duplicate detection, payee re-screening before money moves, payment runs with full audit.
Ask your spend a question.
A governed spend cube with natural-language queries, savings pipelines with verification, budgets that actually block.
Meet the agents.
Fifteen named agents in three classes — autonomous where the blast radius is small, collaborative where judgment matters, ambient where vigilance never sleeps.
Autonomous
Run a process end-to-end within policy, escalating only exceptions — negotiation rounds, clean-invoice posting, screening blocks.
Collaborative
Prepare and recommend — structured intake, contract analysis, supplier briefings. You decide.
Ambient
Watch continuously — risk feeds, renewals, stock levels, spend patterns. They wake you only when it matters.
- Negotiation Agentexecuting
round 3/6 — counter at $8,140, walk-away $8,600
⚡ claude-sonnet2,214 tok$0.04 - P2P Agentexecuted
invoice INV-2214 matched 3-way — auto-posted
⚡ gpt-4o-mini912 tok$0.01 - Screening Agentexecuted
supplier Meridian GmbH — no hits, cleared
⚡ claude-haiku640 tok$0.01 - Approval Agentexecuted
laptop refresh $4.2k — in policy, auto-approved (L2 band)
sim— tok$0.00 - Renewal Agenthandoff
CRM renewal in 41 days — launched renew-vs-resource
⚡ claude-sonnet1,380 tok$0.02
The Autonomy Ladder
One dial, seven stops, enforced per action — the product's own language:
Autonomy with brakes.
Every agent carries its own default and ceiling — Negotiation runs L4, Contract Analyst starts at L1.
Any agent, any tool — paused in one click. Daily run caps and cost budgets enforce restraint.
Every run shows its model, tokens and real dollar cost. Simulated output is always labeled sim.
Built so you can prove it.
Four named primitives make agent autonomy auditable — not a promise, a mechanism.
Action Tokens
Agents cannot write to your systems of record without a workflow-issued token. No token, no write — enforced in the write path.
Tamper-evident Audit
Every action lands in an append-only, hash-chained audit log per tenant — searchable and exportable.
Agent Evals
A golden-dataset regression gate scores every agent, re-runnable live from the console. Agents that grade themselves.
Run Traces
Every agent run is a URL: plan, tool calls, confidence, tokens and cost. Trust what you can trace.
The platform demos itself.
Seven self-driving walkthroughs drive the real product — real routes, real clicks, real agents, honest model and cost chips. Nothing staged.
The flagship: a full day of procurement where a human decides exactly twice.
Supplier discovery
▶ ~5 minThe agent searches the open web, runs outreach, reads replies, recommends. You convert.
Autonomous negotiation
▶ ~4 minA bounded haggle inside human guardrails that banks real savings.
Sourcing RFx
▶ ~4 minA competitive bid scored by the agent — price, delivery, quality, risk.
TPRM sentinel
▶ ~3 minThe risk agent that never sleeps: sanctions, PEP, adverse media, 4th parties.
Fleet evals
▶ ~3 minAgents graded against golden datasets, re-run live. Green means trusted.
Touchless 3-way match
▶ ~4 minA clean invoice clears itself; a price variance is caught and explained.
See supplyce on your own spend.
Tell us about your procurement flow — we'll run a real request through the agent fleet and show you every decision it makes.
- A real request run through the agent fleet, live
- Every agent decision explained — autonomy, cost, audit
- We read everything and reply fast
OR WRITE TO CONTACT@VERTICALISE.COM