What problem does it solve?
Most large buyers negotiate with their strategic suppliers and leave the rest on standard terms. An HBR article co written by Walmart International's sourcing leaders reported that around 20% of Walmart's suppliers had signed standard terms that were often not negotiated, because buyers lacked the time. Pactum's Honeywell story describes hundreds of supplier contracts that were too time consuming to negotiate by hand, and SUEZ recycling and recovery UK says its program reached suppliers beyond its Tier 1 strategic partners.
A tool that only recommends still leaves a person to write to each supplier, wait for the answer and key the result into the system. An agent holds the conversation itself, inside a mandate the buyer writes, and records the outcome. That mandate sets what the agent may offer, what it may never concede and which suppliers it may not contact.
How does it work?
- Select the opportunity. Procurement picks a supplier group and a lever, for example standardizing payment terms, a flat rebate or an item level discount, using spend and contract data.
- Set the mandate. The buyer writes down targets, walk away values, tradable terms and the suppliers or categories that are excluded. Anything outside the mandate stops the agent.
- Open the conversation. The agent contacts each supplier through a chat interface and explains the proposal.
- Negotiate within limits. The agent weighs the supplier's counter offers against the mandate, offers trade offs the buyer allows (for example faster payment for a discount) and closes when terms fall inside the range.
- Write the result back. Agreed terms go to the contract or payment system, and exceptions or refused offers go to a buyer with the conversation attached.
- Keep a record. Every message and concession is logged, so a buyer or auditor can read exactly how a deal was reached.
- Audience
- Back office
- Autonomy
- Supervised agent
- Adoption
- Early adopters
- Channels
- Web chat, Internal tools
What is it worth?
Benchmarks are computed from the public deployments below: one data point per organization per KPI, with who made each claim.
| KPI | Median | Reported range | Data points | Claimed by |
|---|---|---|---|---|
| Interactions handled | Not pooled | at least 2500 | 1 | 1 vendor |
Value drivers: Lower cost to serve, Employee productivity, Speed and cycle time.
Indicative value
A manufacturer with 40 million USD a year of long tail supplier spend
USD 80,000 to USD 600,000
Annual gain on long tail supplier spend per year
How this is calculated
Formula: tailSpend * shareNegotiated * termGain. The low scenario uses every low input, the high scenario every high input.
| Input | Low | High | Basis |
|---|---|---|---|
| Annual spend with suppliers that buyers do not negotiate today tailSpend, USD per year | 40,000,000 | 40,000,000 | The reference organization. Replace with your own spend cube. |
| Share of that spend on which the agent agrees new terms in a year shareNegotiated, share as a fraction | 0.2 | 0.5 | Editorial assumption. Pactum reports that 80% of suppliers contacted at Honeywell engaged, and Walmart International's sourcing leaders reported in HBR that the chatbot closed agreements with 68% of suppliers approached, but those rates cover suppliers selected for the program, so the range is set well below them. Source |
| Average gain on the negotiated spend termGain, share as a fraction | 0.01 | 0.03 | Editorial assumption set at the low end of the 1 to 7% range that Pactum states for its customers on its clients page. SUEZ recycling and recovery UK reports 2.5% average potential savings from targeted discounts. Vendor and customer claims, so the range is conservative. Source |
What it leaves out: Gross gain on negotiated terms only. It leaves out the license and integration cost, the value of payment term changes counted as cash rather than price, supplier relationship effects and the buyer time still spent on exceptions.
Who already uses it?
4 public deployments, strongest evidence first. Grades: A regulator or audit, B the organization itself, C vendor case study, D anonymous or estimate.
SUEZ recycling and recovery UK
United Kingdom · Energy and utilities · 2025
SUEZ recycling and recovery UK uses Pactum's negotiation platform, built into its procurement system, to negotiate prices and payment terms with suppliers through a chatbot interface. SUEZ says the platform reached suppliers beyond its Tier 1 strategic partners, contacted 2,000 additional suppliers within two months and secured discounts through early payment incentives. It reports 2.5% average potential savings through targeted discounts and 15% cost reductions through a competitive purchasing environment. The page is published by SUEZ itself, with a quote from its Head of Procurement.
No outcome disclosed.
Walmart
United States · Retail and ecommerce · 2022
Walmart uses a chatbot from Pactum to negotiate payment terms and price discounts with tail end suppliers, where buyers lack time to negotiate and around 20% of suppliers had signed standard terms that are often not negotiated. The HBR article describing it is co written by two sourcing leaders at Walmart International and two University of Arkansas professors.
Walmart decides the acceptable negotiation trade offs and the bot negotiates and closes agreements within them; the article advises starting in indirect spend categories with pre approved suppliers and scaling by geography, category and use case. The authors report that, so far, the chatbot has closed agreements with 68% of suppliers approached. Pactum, the vendor, reports a 3% average gain across negotiations while extending payment terms by an average of 35 days.
The first is a supplier agreement rate and the second a commercial gain on negotiated terms; neither measures how much of the process runs without human touch or what procurement costs to run.
No outcome disclosed.
Ahold Delhaize USA
United States · Retail and ecommerce · 2026
Ahold Delhaize USA's Indirect Center of Excellence is piloting autonomous supplier negotiations with Pactum, with the aim of cost savings. At DPW New York its Director said the team is piloting with select suppliers and keeps full oversight of each negotiation. No figures were disclosed. Recorded as a pilot: it shows the cautious, supervised way a large buyer starts.
No outcome disclosed.
Honeywell
United States · Manufacturing · 2025
Honeywell started with an indirect spend pilot of Pactum's negotiation agents and then made them its standard route for negotiating with a large supplier base, covering payment term standardization, rebates and item level discounts. Pactum's customer story reports more than 2,500 autonomous negotiations and that 80% of suppliers contacted engaged. The story is written and published by the vendor, with a quote from Honeywell's Chief Procurement Officer.
- Interactions handled: at least 2500, to date, undated story
"They built it into how they operate, deploying nearly half a billion dollars in spend through Pactum AI and concluding more than 2,500 autonomous negotiations to date."
Claimed by: vendor
How do you implement it?
A model agnostic playbook: what to prepare, the order to build in, and what goes wrong.
Data you need
- Spend data by supplier and category, with current prices and payment terms
- Contract register with renewal dates and any exclusivity or notice clauses
- Supplier master with named contacts and language preferences
- Written negotiation mandate per category: targets, walk away values, allowed trade offs
Systems to integrate
- Procurement or source to pay platform (for example Coupa or SAP Ariba)
- ERP for payment terms and contract updates
- Email or a hosted chat page for supplier contact
- Contract lifecycle management system for the agreed terms
Complexity: Medium
A pilot needs a clean spend and contract extract for one category, a purchasing system that can take the agreed terms back, and a written mandate signed off by the category owner and legal. Rolling out across categories needs a supplier master with reliable contact data and a process for suppliers who want to talk to a person.
- 1
Start with indirect spend and a single lever
Pick one category with many small suppliers and one lever, such as standard payment terms. Pactum's Honeywell story says that program began with an indirect spend pilot, and Ahold Delhaize USA runs its pilot in its Indirect Center of Excellence with select suppliers.
- 2
Write the mandate before the agent
Agree the targets, the limits you will not cross, the terms you can trade and the exclusions with the category owner and legal. The agent reads this mandate, so ambiguity in it becomes ambiguity in the deals.
- 3
Run supervised first
Let buyers read every conversation at the start. Ahold Delhaize USA says it keeps full oversight of each negotiation and interaction today, and that the level of review can evolve as confidence builds and governance matures.
- 4
Write results back to the system of record
Connect the agent to the purchasing or contract system so agreed terms are updated by the workflow, not retyped.
- 5
Widen by category and supplier tier
Add categories only after the sampled review shows deals landing inside the mandate.
Guardrails
- The agent can only offer terms inside the written mandate and stops on anything outside it
- Strategic suppliers and suppliers in dispute are excluded from contact
- The agent states that it is an AI system and a person can be requested at any point
- Every message and concession is logged and attached to the supplier record
- Contract changes above a value threshold need a named human approval
KPIs to instrument
- Share of approached suppliers who reply, agree and decline
- Average gain per closed negotiation against the stated baseline terms
- Share of negotiations closed with no buyer touch
- Days from first contact to agreed terms
- Supplier satisfaction after the conversation
Human in the loop
A buyer owns each category mandate and reviews exceptions, refused offers and a monthly sample of closed deals. Suppliers can ask for a person at any time, and a negotiation the agent cannot close within the mandate goes to the buyer with the full conversation.
Common failure modes
- Mandate too loose
- The agent concedes a term the buyer would have refused. Keep the tradable terms short, test the mandate on a small group first and review closed deals each month.
- Supplier relationship damage
- A supplier feels pushed by a bot. Exclude strategic suppliers, offer a person on request and read supplier replies for friction.
- Savings counted twice
- A payment term gain and a price gain are both booked as savings. Agree with finance how each term converts to value before the first deal closes.
- Bluffing or false statements
- To win a concession the agent states something false, such as an invented competing offer, which creates contract law exposure. Allow approved claims only and run test suites that catch false statements.
- Unchanged systems
- Agreed terms sit in chat logs and never reach the contract or payment system. Make write back part of the pilot scope.
What are the risks and rules?
EU AI Act
Limited risk (transparency)
Article 50(1): suppliers who talk to the agent must be told they are dealing with an AI system unless that is obvious. Negotiating commercial terms between companies is not an Annex III high risk use. Check the design if the agent assigns or evaluates the work of individual contractors (Annex III, point 4).
Rules that apply
Guidance
- AI Risk Management Framework (National Institute of Standards and Technology, North America). Voluntary framework, with a generative AI profile, for governing and measuring the risks of an AI system such as an agent that commits the organization to supplier terms.
Controls to put in place
- Inventory entry for the agent with an accountable head of category
- Version controlled mandates with approval by procurement and legal
- Sampled review of closed negotiations, logged with the result
- Access control and retention rules for supplier contact data
Frequently asked questions
- Is this an agent or a recommendation tool?
It is an agent when it contacts the supplier and closes terms itself. Jim Rahill of Pactum, the vendor, put it this way at DPW New York: the agent has to go out and do the work within the defined guardrails, then update the source systems. Ahold Delhaize USA, speaking in the same session, is still piloting and keeps full oversight of each negotiation and interaction.
- What results have buyers reported?
SUEZ recycling and recovery UK says on its own website that the platform contacted 2,000 additional suppliers within two months, and that Pactum's model delivered 2.5% average potential savings through targeted discounts and 15% cost reductions via a competitive purchasing environment; the page does not say what the 15% is measured against. Pactum reports more than 2,500 autonomous negotiations at Honeywell, and Walmart International's sourcing leaders reported in HBR in 2022 that the chatbot had closed agreements with 68% of suppliers approached. None of these figures is audited, and every deployment on this page runs on Pactum.
- Which spend suits it?
Indirect spend and the long tail, where many small suppliers sit on standard terms and no buyer has time. Strategic suppliers stay with people. Pactum's Honeywell story says the program began with an indirect spend pilot, and the Ahold Delhaize USA pilot runs in its Indirect Center of Excellence.
How to cite this page
Blits.ai AI Use Case Library, "AI agent for autonomous supplier negotiation", last verified 11 November 2026, https://www.blits.ai/ai-use-cases/autonomous-supplier-negotiation-agent. Licensed under CC BY 4.0. Method: how we verify use cases.
Changelog
- 11 November 2026: First published