Welcome to edition 04 of /n Procurement.

Last month I broke down how SAP actually prices its AI detailing what's free in the Base AI tier and where the premium meter starts running.

This month is the natural follow-up: now that you know what SAP's AI costs, here's what you'd actually be buying (hint: AI Agents!)

At Sapphire 2026, SAP put a name on the vision: an autonomous suite for the autonomous enterprise. For spend management, that means 11 Joule Assistants and a small army of underlying agents promising to run Source-to-Pay for you.

Which forces the real question:

"Is any of this actually real, and can I trust it to touch my spend?"

Every procurement team on SAP is going to be asked to have an opinion on this soon…

Three important takeaways:

  • The model is three layers. Joule is the interface, Assistants coordinate a business process, Agents do the narrow tasks. 11 assistants now cover most of Source-to-Pay, plus travel, expense, and services.

  • This is bigger than Next-Gen Ariba. The vision spans the whole SAP spend estate: Ariba, Fieldglass, Concur, Business Network, and S/4HANA. Whether it works consistently across all of them is the open question.

  • "Autonomous" is not a switch you turn on. It's an operating model that only holds up on trusted data, clear decision rights, and clean integrations. Put AI agents on a messy foundation and you don't fix the mess… you just let it run faster.

As for whether these assistants are ready to run in the real world? That comes down to six questions… and I walk through all six today.

Let's get into it.

📢 What you'll find in this edition:

  • 🔗 Worth Bookmarking

  • 🔁 Inside SAP Autonomous Spend: What SAP Promises vs. What It Takes

  • 🔵 Best LinkedIn Post of the Month

  • 💬 Worth Repeating…

  • 🌯 Wrap Up

Note: Some of the content listed above is only available in the email version of this newsletter. Don’t miss out! Sign up for free to get the next edition.

The Monthly Update

Exploring SAP’s Roster of Spend Management AI Agents

For the past couple of years, SAP Joule in procurement has mostly been discussed as a user-experience improvement.

Ask a question, find a document, check a requisition status, summarize information, or reach the right application without navigating multiple menus.

Useful? Certainly. Transformational? Not yet.

SAP’s latest spend management direction is more ambitious. Joule is no longer being positioned only as a conversational layer over SAP applications, but as the entry point to a portfolio of Joule Assistants and Joule Agents designed to coordinate work across business processes.

At Sapphire 2026, SAP moved this narrative further by introducing the forward-looking vision of an autonomous suite for the autonomous enterprise.

Autonomous Enterprise:

For those following SAP’s delayed but accelerating AI push, this shift is not surprising. The question is no longer whether AI can create value; it is how that value becomes embedded into the way the business actually operates.

That requires more than new tools. It requires a different operating model.

An Autonomous Enterprise in its essence is an organization where AI is embedded into the flow of work, connecting signals, decisions, and actions across the business. People set the objectives, policies, and guardrails. AI executes routine decisions and processes at scale.

The difference is significant. When AI is layered on top of existing ways of working, it improves individual tasks. When AI is embedded into the operating model, it transforms how the business functions, enabling faster decisions, greater coordination, and a more agile response to change.

In traditional organizations, systems generate information, people interpret it, and teams coordinate actions. Every handoff introduces delay. In an Autonomous Enterprise, signals are continuously monitored, interpreted in context, and acted upon in real time. A supply disruption can trigger responses across planning, procurement, and operations. A financial risk can initiate corrective actions before it becomes a larger problem. 

Ultimately, an Autonomous Enterprise is not a product or a technology initiative. It is a new way of operating, where AI helps the business move faster, respond smarter, and execute more effectively at scale.

So what does SAP’s framework for this autonomous enterprise look like?

Joule is the engagement layer. Rather than navigating multiple systems and transactions, users express an intent or desired outcome. Joule then brings together the relevant data, workflows, and AI agents across SAP and third-party applications. Through Joule Work, users can direct assistants that coordinate specialized agents to execute work across the enterprise.

SAP Autonomous Suite is the execution layer. This is where autonomous capabilities operate across key business domains such as Finance, Spend, Supply Chain, Human Capital Management, and Customer Experience. Each domain combines AI assistants and agents that can make decisions, trigger actions, and orchestrate processes within defined business guardrails.

Industry AI provides domain and industry-specific intelligence. These capabilities are embedded directly into business processes and reflect the realities of different industries, including their process requirements, regulatory obligations, and operational nuances.

SAP Business AI Platform is the foundation layer. Built on SAP Business Technology Platform, Business Data Cloud, AI Foundation, and Business Transformation Management capabilities, it provides the data, governance, integration, and scalability needed for AI agents to operate reliably and responsibly across the enterprise.

Together, these layers connect user intent, business intelligence, process execution, and governance, creating an operating model where AI is embedded into how work gets done, not simply added as another feature.

SAP Autonomous Suite

AI agents need more than powerful models. They rely on trusted data, well-defined processes, and strong governance. SAP provides these foundations through semantically rich business data, end-to-end process context, and embedded controls across Finance, Supply Chain, Spend, HCM, and CX. This enables AI agents to operate within clear business guardrails while maintaining compliance, auditability, and data integrity.

The SAP Autonomous Suite is the operational layer of this model. It organizes autonomous capabilities across five core business domains, each following the same basic pattern: assistants coordinate work, agents perform specialized tasks, and business context provides the guardrails.

For this article, the focus is autonomous spend: the assistants and agents SAP has outlined for its spend management portfolio over the coming months.

SAP Autonomous Spend

SAP is positioning its autonomous spend management portfolio around the goal of maximizing spend visibility and source-to-pay efficiency through seamlessly orchestrated, self-improving processes. According to SAP, this approach enables organizations to increase productivity, transform savings into measurable business value, and strengthen control over risk, compliance, security, and sustainability.

SAP further states that this vision will be enabled through AI-powered assistants and agents capable of automating strategic sourcing activities, supporting intelligent contract creation, guiding compliant purchasing decisions, streamlining accounts payable operations, and driving continuous improvement across the source-to-pay lifecycle.

Customers, however, should view this vision with a fair degree of pragmatism. An AI assistant and agent-led future realizing the fact that meaningful value will require sustained executive sponsorship, high-quality and well-governed data, standardized processes, and a broader commitment to cloud-native intelligent platforms and connected business workflows. Organizations that invest in these foundational capabilities will be significantly better positioned to capture the value SAP is promising.

The Autonomous Spend portfolio spans five areas, each powered by Joule Assistants and Agents that orchestrate end-to-end processes:

Together, these assistants cover much of the source-to-pay lifecycle, alongside travel, expense, and external workforce processes. 

SAP’s AI terminology can become confusing very quickly, so first, Let Us Fix the Terminology

Joule. Joule Assistants. Joule Agents. These concepts are related, but they are not interchangeable.

Joule is SAP’s conversational AI experience. It provides an interface through which users can ask questions, access information, receive recommendations, and initiate supported actions.

A Joule Assistant is aligned with a broader business process or outcome, such as sourcing, supplier management, contracting, or invoicing.

A Joule Agent is a specialist component designed to perform a more narrowly defined job within that process.

The simplest way to understand the model is: Joule is the interface. The assistant coordinates the work. The agents perform specialized tasks.                                                                                     

Here are the assistants and underlying agents that SAP has promised its customers in the source-to-pay space. Do keep in mind SAP keeps updating the information and delivery timelines, so refer SAP’s official documentation for the latest.

1. Category Management Assistant

The Category Management Assistant is intended to improve spend classification, surface risks and opportunities, and support category strategy development as market and business conditions change.

SAP lists four agents underneath it:

  • Spend Analysis Agent

  • Category Strategy Development Agent

  • Market Insights Agent

  • Spend Opportunity Agent

It is expected to combine spend, supplier, contract, forecast, and market inputs to support more continuous category optimization.

The promise depends on trustworthy spend classification, supplier hierarchies, contract links, demand signals, and external market data.

2. Sourcing Assistant

The Sourcing Assistant is positioned to support the sourcing lifecycle from supplier discovery and event creation to bid analysis, award recommendations, negotiations, and counteroffers.

SAP currently lists:

  • Sourcing Event Agent

  • Bid Analysis Agent

  • Sourcing Negotiation Agent

SAP says it can use past events, contracts, supplier records, risk inputs, market data, and sourcing objectives to prepare events and evaluate bids.

The key requirement is explainability: buyers must understand the criteria, weighting, data sources, assumptions, and policy logic behind any recommendation.

Supplier and Contract Management: From Records to Continuous Monitoring

3. Supplier Management Assistant

The Supplier Management Assistant spans supplier discovery, creation, enrichment, classification, due diligence, risk monitoring, performance assessment, corrective actions, and lifecycle management.

Its specialist agents include:

  • Supplier Management Creation Agent

  • Supplier Management Classification Agent

  • Supplier Management Due Diligence Agent

  • Supplier Management Risk Monitoring and Alert Agent

  • Supplier Management Identification and Optimization Agent 

The value is clear because supplier data is often fragmented across entities, qualification records, risk providers, and performance processes.

The more consequential the supplier decision, the more important human review, trusted sources, and complete audit trails become.

4. Procurement Contract Assistant

The Procurement Contract Assistant supports contract authoring, analysis, compliance monitoring, renewal planning, and contract-related mass changes.

SAP lists agents relating to:

  • Contract authoring

  • Contract analysis

  • Compliance monitoring

  • Renewal optimization

  • Purchase contract mass changes

  • Purchasing information record mass changes

  • Material spend advisory

SAP describes agents that extract contract data, monitor commitments, detect leakage, recommend renewals, and support approved updates to purchasing records.

Customers will need clear controls distinguishing what an agent can analyze, recommend, prepare, submit for approval, or execute.

Operational Buying: Making Procurement Complexity Invisible

5. Requisition Assistant

The Requisition Assistant helps users start the purchasing journey by guiding demand to the right channel, supplier, commodity, account assignment, and approval path.

The supporting agents are:

  • Purchase Requisition Creation Agent

  • Purchase Requisition Optimization Agent

  • Demand Aggregation Agent 

Its real value is reducing requester complexity while applying procurement policy in the background.

The opportunity is not just faster requisition creation, but fewer unnecessary steps with controls still intact.

6. Buying Assistant

The Buying Assistant focuses on the commercial route after demand is identified: catalog quality, source assignment, supplier recommendations, and contract coverage.

SAP lists:

  • Material Catalog Management Agent

  • Supplier Recommendation Agent

It is expected to normalize catalog content, identify purchase orders without contract assignment, and recommend suppliers using purchase history, performance, pricing, and contract data.

Because requisition and buying experiences will overlap, customers should clarify which application owns the data, recommendation, workflow, and transaction.

Receiving and Invoicing: Fixing Problems Before They Reach Accounts Payable

7. Receiving Assistant

The Receiving Assistant supports goods receipts, service entry sheets, quality tracking, and returns.

SAP lists:

  • Goods Receipt Creation Agent

  • Service Entry Creation Agent

  • Quality and Returns Agent

  • Return Goods to Supplier Agent 

Its value is preventing missing or incorrect receipts from becoming blocked invoices, delayed payments, inaccurate accruals, and avoidable follow-up work.

The real value is preventing avoidable problems from reaching accounts payable.

8. Invoicing Assistant

The Invoicing Assistant targets invoice capture, validation, exception handling, coding, fraud checks, tax validation, contract matching, approval routing, and posting.

The ambition is close to autonomous invoice processing, but touchless invoicing still depends on clean POs, timely receipts, reliable supplier data, correct tax logic, stable integrations, and governed exceptions.

The agent may become the visible intelligence. The invisible foundations will still determine whether the invoice posts or fails.

Services, Travel, and Expense: Extending the Model Beyond Ariba

9. Services Procurement Assistant

The Services Procurement Assistant supports the statement-of-work lifecycle, from SOW creation and amendments to milestone tracking, acceptance evidence, insights, and market-rate guidance.

Its listed agents are:

  • SOW Creation Agent

  • SOW Change Management Agent

  • SOW Insights Agent

  • SOW Workforce Market Rate Agent

This matters because services procurement depends on clear deliverables, measurable outcomes, evidence of acceptance, controlled changes, and invoice validation against actual delivery.

A well-written SOW is useful, but evidence, ownership, and formal acceptance still determine whether delivery can be trusted.

10. Travel Assistant

The Travel Assistant supports individual and group travel planning through cost estimates, policy guidance, itinerary generation, booking support, approvals, and meeting logistics.

The listed agents are:

  • Booking Agent

  • Meeting Planning Agent 

It extends the spend story into SAP Concur and reinforces why SAP is organizing Joule around business journeys, not only individual applications.

The practical challenge is coordinating preferences, policy, budgets, approvals, booking content, meeting locations, and downstream expense processing.

11. Expense Management Assistant

The Expense Management Assistant supports the journey from receipt capture to expense report completion, validation, policy guidance, and audit-rule support.

Its supporting agents include:

  • Receipt Analysis Agent

  • Expense Automation Agent

  • Expense Report Validation Agent 

Employees will appreciate faster expense reporting, but the bigger value is fewer errors, missing receipts, approvals delays, and downstream finance rework.

The Underlying Architecture Will Determine the Outcome

The outcome will depend less on the assistant label and more on the architecture underneath it: SAP BTP, identity, integration, provisioning, application services, and governed business data.

Customers should therefore evaluate these assistants against the foundations that will determine whether they can operate reliably:

  • Identity and authorization

  • Clean and connected business data

  • Consistent domain models

  • Application APIs

  • Workflow orchestration

  • ERP and third-party integration

  • Policy and approval rules

  • Monitoring and auditability

  • Reliable master and transactional data

This is why a controlled product demonstration is not enough.

The real test is whether an agent can function inside fragmented data, customized approval structures, regional requirements, legacy integrations, and established governance models.

What Customers Should Ask SAP

Before treating these assistants as production-ready building blocks, procurement and IT leaders should ask six practical questions.

1. Availability

  • Which assistants and agents are generally available today?

  • Which remain roadmap concepts, design validations, or limited-release capabilities?

  • Which are available in current-generation SAP applications?

  • Does availability vary by application, region, or data center?

2. Licensing and consumption

  • Which capabilities are included in existing subscriptions?

  • Which require separate AI entitlements or consumption units?

  • What activity constitutes a billable AI transaction?

  • How can customers forecast consumption?

  • Can administrators define limits by agent, process, user group, or business unit?

3. Data grounding

  • Which SAP and non-SAP data sources can each agent use?

  • How current is the underlying information?

  • Can users see which sources contributed to a recommendation?

  • How does the agent handle conflicting or incomplete data?

  • Can customers exclude selected information sources?

4. Decision authority

  • Does the agent inform, recommend, prepare, submit, approve, or execute?

  • Which actions require human confirmation?

  • Can approval and execution thresholds be configured?

  • How are segregation-of-duties requirements enforced?

  • What happens when an agent recommendation conflicts with an established workflow rule?

5. Explainability and auditability

  • Can users see why a recommendation was made?

  • Are the relevant inputs, outputs, confirmations, and actions logged?

  • Can an organization reconstruct an agent-supported decision later?

  • Can an incorrect recommendation be challenged and corrected?

  • How long are agent activity records retained?

6. Extension and support

  • Can customers configure or extend SAP-delivered agents?

  • Can organization-specific policies and rules be added?

  • How will customer-built agents coexist with SAP-delivered agents?

  • How are responsibilities divided between SAP, implementation partners, and customers?

  • Who supports a workflow when multiple SAP and custom agents participate?

These questions are not technical detail.

They determine whether SAP’s autonomous spend vision can become a controlled, scalable enterprise capability.

My Take

SAP’s direction makes sense because procurement work remains fragmented across too many applications, roles, documents, approvals, and sources of information.

Users should not need to understand the entire system landscape to buy an item, compare suppliers, review a contract, post a receipt, or resolve an invoice exception.

Process-aware assistants coordinating specialist agents are a more credible path than one general-purpose chatbot attempting to do everything.

But customers should be careful with the word autonomous.

Autonomy is not a feature to switch on; it is an operating model built on trusted data, clear decision rights, reliable integrations, explainable recommendations, security controls, accountable ownership, and complete audit trails.

Most organizations are not starting from a clean slate; they already carry inconsistent supplier data, incomplete contract coverage, customized workflows, regional policy variations, missing receipts, integration failures, and years of accumulated exceptions.

Putting agents on top of that landscape will not automatically resolve the complexity; in some cases, it may simply allow the complexity to operate faster.

The strongest part of SAP’s direction is the process-oriented assistant model: assistants that coordinate specialized agents around business outcomes rather than disconnected AI features inside individual applications.

The unanswered question is how consistently this model will work across current-generation SAP applications, next-generation SAP Ariba, SAP S/4HANA, SAP Business Network, SAP Fieldglass, SAP Concur, and customers’ non-SAP systems.

That is where the real work begins.

The Takeaway for Procurement Pros

SAP’s Joule strategy for spend management is becoming more substantial.

It is moving beyond status checks and screen navigation toward a model where Joule is the entry point, assistants coordinate processes, agents perform defined tasks, and business data grounds decisions.

Organizations evaluating these capabilities should start by mapping assistants to real process pain points, then assessing whether the required data, controls, integrations, ownership, skills, and commercial model are ready to support them.

Do not begin with:

Where can we deploy an AI agent?

Begin with:

Which business decision or process failure are we trying to improve, and do we trust the foundations on which the agent will operate?

Because autonomous procurement will not be achieved by asking AI to compensate for years of unresolved process and data problems. The future may be agentic. But trust will still be built through reliable execution, transparent decisions, and fixing the foundations underneath.

Disclosure: This article represents the understanding and views of the author at the time of writing. /n Procurement is an independent publication and is not affiliated with, endorsed by, or sponsored by SAP. Readers should independently validate all information provided before making any major business decisions. Use at your own risk.

Worth Repeating…

Automation applied to an inefficient operation will magnify the inefficiency.

Bill Gates

Wrap Up

  1. The promise of AI in procurement gets a lot of attention.
    The real test is whether people trust it enough to use it. In this discussion, ConvergentIS gets practical about adoption, process clarity, and using AI for insight without pretending it can do everything.

  2. Real procurement transformation takes more than new technology.

    This session brings together Paula Glickenhaus, Emily Rakowski, Joël Collin-Demers, and Ian Lawless to break down how leading teams define scope, align sponsors, build the business case, and turn complexity into progress.

The /n Procurement Team

P.S. Please rate today's newsletter.

Your feedback shapes the next edition.

Login or Subscribe to participate

Reply

Avatar

or to participate