Mor Cohen-Tal

How Much Autonomy Should Procurement AI Agents Have?

AI agents trust ladder

If you lead procurement, you have probably been asked some version of this question in the last year: how much of this can the AI just do? It is a fair question, and the honest answer is that it depends on the workflow, not on the AI. The teams getting value from procurement AI agents decide autonomy one workflow at a time, and they move up only when the results earn it.

We call that approach the trust ladder. It has three rungs: manual, rule-based and hands-off. This post explains each rung, how to decide where a workflow belongs and which controls should stay in place at every level.

Mor Cohen-Tal, Co-Founder and CTO of Opstream
By Mor Cohen-Tal, Co-Founder and CTO
Co-Founder and CTO of Opstream, previously Cloud CTO at Turbonomic (acq. IBM for nearly $2B) and holds 13 patents in cloud and AI infrastructure.
View LinkedIn profile →
67%
of procurement organizations are discussing or planning to implement agentic AI
Gartner, Top Insights on AI for Chief Procurement Officers, May 29, 2026
33%
are dissatisfied with the ease of implementation
Same survey
21%
worry about the rework it would introduce
Same survey

Why is autonomy the real question for procurement AI agents?

Autonomy decides who carries the risk when something goes wrong. A procurement AI agent that drafts a summary carries almost none. One that approves a contract carries a lot. The technology is the same. The consequence is not.

Gartner said in July 2025 that generative AI for procurement had entered the Trough of Disillusionment, with many organizations seeing uneven ROI. The obstacles it listed were fragmented, low-quality data, complex integration and skepticism about AI-driven insights. Those are trust problems as much as technology problems.

According to the 2025 Gartner Procurement Organization Design for the Future Survey, 67% of procurement organizations are discussing or planning to implement agentic AI. However, 33% are dissatisfied with the ease of implementation, and 21% worry about the rework it would introduce.

Interest is high and confidence is not. The gap closes when teams can see what an agent did, why it did it and how to stop it.

What are the three rungs of the trust ladder?

Each rung gives the workflow more room to act on its own. Each one also keeps a clear place for people.

RungWhat runsGood fit
ManualPeople review, decide and act. The platform records each step and suggests answers marked for review.New workflows, rare cases, anything with poor data
Rule-basedConditions, approval brackets, assign-only-if rules and reminders route the work the same way every time.Policy you can write down, such as approval limits and renewal reminders 90 days before renewal
Hands-offAgent steps run when the step is reached, decide what they found and route on the result. People handle exceptions.High-volume checks with clear outcomes, such as contract compare and playbook checks

Rung 1: Manual

People do the work, and the platform makes it visible. Requests, answers, approvals and documents live in one place. AI helps without deciding: Opstream can suggest answers from an uploaded quote, and those values stay marked as suggestions until a person accepts them. Ask Opstream answers questions about requests and vendors, and it shows which tools it is using while it answers.

Rung 2: Rule-based

You write down the decisions people repeat, and the workflow follows them. Approval brackets set who signs at each amount. Conditions add Legal or Security only when the request needs them. Reminders start a renewal review a set number of days before the contract ends. Rules are predictable, which is why most procurement policy belongs here.

Rung 3: Hands-off

Agent steps run when the workflow reaches them and route on what they find. In Opstream, a Compare step picks the documents to compare, such as a renewal against the previous agreement, and reports changes by severity. A Playbook step checks a contract against your organization’s positions for that document type. The step outcome drives the next step: high severity goes to Legal, none continues.

How do you decide which rung a workflow belongs on?

Score each workflow on three questions. The answers tell you where to start.

  • How clean is the data? An agent can only be as right as the record it reads. If vendor data conflicts across systems, fix the data model first.
  • How reversible is the outcome? Routing a request to the wrong reviewer is easy to fix. Signing the wrong contract is not.
  • How clear is the rule? If two reviewers would decide the same case differently, the workflow is not ready for hands-off.

A workflow with clean data, reversible outcomes and a clear rule can move up fast. A workflow that fails any one of the three should stay lower until it passes. This is why the data model comes before autonomy, not after it.

Which controls should stay in place at every rung?

Some controls never go away, however much a workflow runs on its own.

Agents are digital workers with decisioning + execution capability. They need identity, permissions, supervision and accountability.

Source: The Hackett Group, “Agentic Governance and Controls Primer,” Kyle McNabb, Ammar Mohammed, June 2026.

  • Show the reasoning. The Auditor step explains each discrepancy and cites the file or answer it relied on. Pre-filled vendor questionnaire answers show their source.
  • Let people correct the input. If an agent picked the wrong documents, users change the selection and re-run.
  • Respect permissions. Ask Opstream and the MCP server only show what the current user is allowed to see.
  • Log everything. Audit logs record each action and can be filtered and exported.
  • Keep advisory agents advisory. The Auditor step never edits request data. It suggests the corrected value and leaves the change to a person.

How do you move a workflow up the trust ladder?

Move one rung at a time and measure before each move. These six steps work for most procurement teams:

  1. Pick one workflow with clean data. Start where the data is complete and the rules are written down, such as software requests or renewals.
  2. Run it manually in the platform first. Let people do the work in Opstream so every step, answer and decision is recorded.
  3. Write the rules you already follow. Turn the decisions people repeat into conditions, approval brackets and reminders.
  4. Add an advisory agent step. Add an Auditor or Compare step that reports findings but does not decide. Check its findings against what reviewers decide.
  5. Let the agent route low-risk outcomes. When findings match reviewers, let the step route on its result: none continues, high goes to a person.
  6. Review the audit trail every month. Check exceptions, overrides and re-runs. Move the workflow up or down a rung based on what you see.

The trust ladder applies to every procurement use case. Contract review often reaches hands-off first, because a comparison has a clear result.Third-party risk at intake usually stays rule-based, with a person approving every high-risk vendor.

Can you bring your own AI to procurement?

Yes, and it fits the same ladder. The Opstream MCP server connects an AI assistant your company already uses to your Opstream data. It answers within each user’s permissions. When a request is created through it, the same checks run as in the app, including the check for software you already own. Your assistant gets context, and your controls stay where they are.

Key Takeaways
• Decide autonomy per workflow, not per platform.
• Manual, rule-based and hands-off each keep a clear place for people.
• Clean data, reversible outcomes and clear rules decide how high a workflow can go.
• Show the reasoning, allow re-runs, respect permissions and log every action at every rung.
• Move up one rung at a time, and measure before each move.

FAQ: Procurement AI Agents and Autonomy

What is the trust ladder for procurement AI agents?

The trust ladder is a way to decide how much a procurement workflow runs on its own. It has three rungs. Manual means people do the work and the platform records it. Rule-based means conditions you set route the work. Hands-off means agent steps run and route on their own result, with people handling exceptions.

Should procurement AI agents approve purchases on their own?

Only for low-risk, well-defined cases with clean data and written rules, such as a renewal with no material changes under a set amount. High-risk vendors, large contracts and policy exceptions should always reach a person. Opstream lets admins choose which steps can auto-approve and which always need a reviewer.

How do you keep AI agents in procurement auditable?

Give every agent step a clear scope, show the evidence behind each finding and log every action. In Opstream, agent steps show their findings on the task card with a link to the full report, users can change the inputs and re-run, and audit logs record who or what did each action.

What is the difference between rule-based automation and agentic AI?

Rule-based automation follows conditions a person wrote, such as routing requests over a set amount to Finance. Agentic AI reads the request and its documents, decides what it found and acts on that, such as comparing a renewal with the previous contract and routing on severity. Most teams need both.

Can we use our own AI assistant with Opstream?

Yes. The Opstream MCP server lets an AI assistant your company already uses query your Opstream data, such as requests, vendors, spend and documents, within each user’s permissions. Requests created that way run the same checks, including the duplicate software check, as requests created in the app.

Where should a procurement team start with AI agents?

Start with an advisory step on a workflow you already run in the platform. The Auditor step is a good first choice because it only suggests corrections and never edits request data. Compare its findings with what reviewers decide for a few weeks, then let it route the clear cases.

See the trust ladder on your own workflows

We will map one of your workflows to the three rungs and show where agent steps fit today.

Book a Meeting
About the Author
Mor Cohen-Tal, Co-Founder and CTO of Opstream
Mor Cohen-Tal
Co-Founder and CTO, Opstream

Mor Cohen-Tal is a visionary technology leader and the Co-Founder and Chief Technology Officer of Opstream, an intelligent procurement orchestration platform that is transforming the way companies buy.

With a career marked by a relentless pursuit of innovation, Mor has earned 13 patents for her groundbreaking work. Notably, Mor was the Cloud CTO at Turbonomic, where she spearheaded the company’s successful transition from a datacenter-focused business to a cloud-centric model. Turbonomic was acquired by IBM for nearly $2B in 2021.

As a leading thought leader in cloud and AI, Mor plays a critical role in cultivating partnerships with leading cloud providers such as AWS and Microsoft Azure, and has presented and keynoted at conferences around the world, including Microsoft Ignite and AWS re:Invent.

Mor holds an M.Eng from Cornell University and a B.Sc from the Hebrew University.

Connect with Mor on LinkedIn →

Sources

  1. Gartner: Gartner Says Generative AI for Procurement Has Entered the Trough of Disillusionment (press release, July 30, 2025)
  2. Gartner, “Top Insights on AI for Chief Procurement Officers,” Micky Keck, Magnus Bergfors, 29 May 2026. GARTNER is a registered trademark and service mark of Gartner, Inc. and/or its affiliates in the U.S. and internationally and is used herein with permission. All rights reserved.
  3. The Hackett Group, “Agentic Governance and Controls Primer,” Kyle McNabb, Ammar Mohammed, June 2026.

Want to see how it works?

Book a demo with our team or reach out at support@opstream.ai