Skip to main content

SAP Connect 2026: The Autonomous Enterprise Clocks In, CX Gets a Cameo

SAP used its Las Vegas keynote to put the autonomous enterprise into general availability. Customer experience got two minutes on stage and found its real airtime in the executive Q&A.

The News

Five months after introducing the autonomous enterprise at Sapphire under the tagline the beginning of better, SAP CEO Christian Klein opened the SAP Connect keynote in Las Vegas on October 6 by showing it in action. He closed with an upgrade of the tagline: “This is no longer the start of something better. This is better.”

Klein’s analogy was a unicorn. Frontier models can now draw one with the correct number of ears, he said, but they still cannot tell you which of your suppliers will deliver next week. In business, “impressive is not enough. AI has to be accurate.” SAP’s answer to this bases on three pillars.

Joule Work is the AI engagement layer and, in Klein’s words, a new UI for AI. It rolls out to all customers starting this month and is built on conversations, spaces for visualizing SAP and non-SAP content, and Joule Studio for extending SAP agents with n8n or building new ones. SAP says Joule Work is live for its 110,000 employees, with productivity gains above 20% in HR, finance and procurement; Klein showed it cutting his earnings call preparation from three weeks to days. It is open in both directions: other AI systems can call Joule through SAP’s agent-to-agent gateway, and Joule can call third-party agents registered in the SAP AI Agent Hub.

The SAP Autonomous Suite spans the five domains finance, spend, HCM, supply chain and customer experience, plus Industry AI. Assistants orchestrate agents that run inside SAP’s deterministic workflows, so that, according to Klein, they deliver the one right answer, follow the same rules and leave the same audit trail as the applications. Customers choose the level of autonomy, from human-in-the-loop to fully autonomous with write-back. Klein said SAP has made more than 20 assistants and over 200 agents generally available since Sapphire and aims for more than 400 agents by year end.

The SAP Business AI Platform is where customers build, contextualize and govern. Klein’s rhetorical question is “Why would anyone build agents standalone on an LLM platform?”

Manoj Swaminathan, President and Chief Product Officer of the SAP Autonomous Suite, demonstrated a merger onboarding scenario that combined SAP and non-SAP HR data and turned a business user’s request into an agent extension. Jan Gilg, SAP’s Chief Revenue Officer for the Americas and SAP Business Suite, then built a fictional company, Jerseys from Jersey, with customer guests. Novartis described a sourcing agent pilot from which it expects 25% faster cycle times and a 35% productivity uplift. Morgan Foods said S/4HANA and IBP cut its response time to large customer requests from three or four weeks to about five minutes. Nestlé presented accounts receivable agents and plans for touchless order management. Gilg also announced the acquisition of TechWolf for SAP SuccessFactors and SAP Pay, a payment service embedded in SAP Cloud ERP.

Entry Customer Experience. Gilg called it the domain “that underpins everything,” used the jersey rush after Lionel Messi’s move to Inter Miami as his example of a CX moment, and said the same data that powers the supply chain also personalizes the customer experience. The guest who followed was PwC’s Chief Accounting Officer Jill Notz, who spoke about industry AI and a billing assistant PwC is building with SAP. A governance segment followed, built on three levels of human oversight: in the loop, on the loop and in command.

In the executive Q&A, Klein named three takeaways: delivery on the Sapphire promise, context (turning 7 million fields and half a million tables into an ontology) and governance. He called APIs and MCP servers “table stakes” and said SAP will now build new application features only where data or compliance requirements are missing and redirect the remaining development capacity to agents. On accountability, he described SAP’s own financial close as moving from roughly 80% automation toward 90 to 95%, with a human signing off and the Big Four auditors working to certify agents.

CX came up three times in the Q&A, each time from the floor. Asked how CX underpins the other pillars, Gilg said front-office experiences fail without back-office connectivity, adding that SAP is “not necessarily the market leader on the CX side, but I think there we can be also very disruptive.” Klein said he had seen a CX vendor demonstrate an agent writing back into an SAP system and believed “that demo will never come true”; third-party agents, he argued, should call SAP through its gateway instead.

When I asked when SAP would take CX out of the Cinderella corner, Klein pointed to heavy investment in contact center solutions, a partnership with Parloa with more partnerships to come, an SAP-built pricing agent for configure-price-quote, and SAP’s own sellers using Joule Work for customer briefings and account planning. He promised the audience would hear more next time. Swaminathan described autonomous CX as agents across lead generation, case management, quoting and the contract lifecycle, and said CX is “absolutely not” being marginalized but receives “huge investment.” Klein closed with his own recent CX sales cycles: in commerce deals, the conversation had moved from availability-to-promise features to whether SAP’s shopping assistant can work out the fastest delivery to the consumer. “Now we play another game.” Asked by Constellation Research’s Liz Miller how CX signals would flow back into supply chain and HR, he pointed to demand planning at Procter & Gamble and to Reltio and Dremio harmonizing customer data across data lakes without forcing customers into a new data strategy.

The Bigger Picture

Autonomous is the buzzword of 2026. ServiceNow launched an Autonomous Workforce at Knowledge 2026 in May, Zendesk an Autonomous Service Workforce at Relate 2026, and Salesforce brought Agentforce Coworker and a back-office agent called Marshall to Dreamforce 2026. Oracle ships agents inside Fusion Applications at no additional charge. With Workday and Oracle holding their own events in the coming weeks, one analyst in the Q&A called it “the October of ERP.”

Each vendor approaches autonomy from its home turf: Salesforce from the interface and intent layer, ServiceNow from governance and workflow, Microsoft from identity and the endpoint, SAP from the process of record. At Connect, SAP contested the other positions with Joule Work as an interface layer for more than 400 million SAP users, the AI Agent Hub as a cross-vendor agent inventory, and Business Data Cloud as the context layer.

Klein’s rebuttal of third-party write-back is a direct jab at Salesforce’s Dreamforce message. Salesforce says Marshall’s Systems Integration Agent executes deterministically in core systems without code, APIs or RPA. SAP’s position is the reverse: the CRM’s agent asks, SAP’s agent transacts, and the audit trail stays in the system of record. This follows SAP’s earlier stated position that you cannot govern what you cannot control.

In customer service, pricing is contested, too. Salesforce charges USD 2 per Help Agent resolution, and Zendesk bills verified resolutions. SAP prices agentic work per action, drawn down from prepaid AI Units.

Within SAP, CX’s weight is measurable. SAP’s analyst summit material counts 15 CX assistants with 62 agents, out of 416 agents across the five domains, and the Connect news guide dates the new and extended CX capabilities to general availability in Q1 2027. At Sapphire, SAP had announced ten named CX assistants for Q3 2026. 15 overachieves on this objective.

My Analysis and Point of View

In general, I think that a follow-up keynote five months after a vision launch needs to show delivery to promise. On the autonomous enterprise as a whole, Tuesday’s keynote largely did. On CX, it did not. Cinderella doesn’t yet dance with her prince.

SAP’s message got simpler. Sapphire was about architecture; Connect was about implementation. Klein’s unicorn metaphor is more important for SAP’s positioning than any agent count: frontier models are good at pictures and bad at your supplier list. The customers on stage were candid about where they stand. Novartis is semi-autonomous and talks about expected gains. Morgan Foods’ answer is a planning story built on S/4HANA and IBP, and its CFO described the company as early in its use of agents, but with impressive results. Bringing a response time down from 3+ weeks to 5 minutes is no mean feat. Klein’s own financial close apparently sits at roughly 80% automation with a human signing off. SAP’s autonomous enterprise is, for now, a supervised enterprise, and SAP said so. I prefer that to the alternative. Customers are not yet there, hence autonomous enterprise with the handbrake pulled is the right answer. And SAP gives it.

The most consequential sentence came in the Q&A. SAP will stop building features where the data and process already exist and put that capacity into agents. For a vendor whose ERP is, in Klein’s words, probably the most feature-rich on the market, this is an R&D reallocation with direct consequences for roadmaps, enhancement requests and user group negotiations. Customers can expect their next enhancement request to land in an agent, not on a screen. Paired with Joule Work replacing four decades of transaction codes, SAP is now making the argument I have made for years: the conversation is the UI.

For CX, the structure of the keynote was the message. Gilg called CX the domain that underpins everything. It then got about two minutes, a hypothetical jersey shop and a football anecdote, and its witness was a chief accountant talking about billing. No CX customer, no CX demo, no CX executive. In fairness, PwC’s billing assistant serves a customer-facing process, and Nestlé’s planned touchless order management came close. That is exactly the Cinderella pattern: CX value shows up inside other domains’ stories while CX itself goes unchampioned. This needs to change for SAP’s end-2-end story to stay intact.

The Q&A revealed the CX strategy the keynote did not tell. Put the answers from Gilg, Klein and Swaminathan together and there is a different picture. SAP CX is meant to be the front end of order-to-cash, from lead to quote to order to contract. SAP builds agents where the transaction lives (pricing, CPQ, order management, commerce) and partners where it does not (voice and contact center, via Parloa). Klein’s sales-cycle anecdote is the best argument for it: if agentic AI moves large-enterprise CX selection from feature checklists to cross-domain outcomes, SAP sheds its historic weakness and plays to its strength. Gilg’s admission that SAP is not the CX market leader is kind of the understatement of the year. Functionally, SAP has what it needs; this was also clearly shown on the show floor. What SAP CX still does not have is executive air cover, aka main keynote time.

The write-back stance cuts both ways. Klein is right that a third-party agent posting into financials and order management needs an audit trail and governance that most CRM platforms lack. But the logical end point of “call us through the gateway” is an SAP that owns the transaction while Salesforce, Microsoft or ServiceNow own the customer conversation. That is a good business. It is not a CX business. If the conversation is the UI, whoever owns the conversation decides which back end gets called, and how often each meter runs. SAP’s vehicle for this is Joule Work. So the company needs to keep Joule Work decluttered while taking advantage of the increasing excess baggage that MS Teams and Slack carry.

In summary, the autonomous enterprise message is an upgrade. It is clearer, more concrete, credibly governed and now rolling out, assuming that general availability comes as promised. That leaves the CX portion. Here’s clear room for improvement. Functionally, SAP has what it takes. The strategy is better than its stage time, and Klein promised more next time. I will hold him to it.

Cinderella got a mention at the ball, but she still came in through the kitchen. 

Comments

Last Year's Top 5 Popular Posts

You are only as good as your customer remembers

As you know, I am very interested in how organizations are using business applications, which problems they do address, and how they review their success. In a next instance of these customer interviews, I had the opportunity to talk with Melissa Gordon , Executive Vice President, Enterprise Solutions at Tidal Basin about their journey with Zoho. You can watch the full interview on YouTube. Tidal Basin is a government contractor that provides various services throughout the government space, including disaster response, technology and financial services, and contact centers. Tidal Basin started with Zoho CRM and was searching for a project management tool in 2019. This was prompted by mainly two drivers. First, employees were asking for tools to help them running their projects. Second, with a focus on organizational growth and bigger projects that involved more people, Tidal Basin wanted to reduce its risk exposure and increase the efficiency of project delivery. This way, the compa...

SAP Draws a Perimeter around Agentic AI and What That Means for the Rest of US

The most consequential enterprise AI governance document published this year arrived in late April with surprisingly little fanfare. SAP's updated API Policy, version 4/2026 , is a short document in plain English. The clause that is most interesting is Section 2.2.2. It restricts how autonomous and generative AI systems are permitted to interact with SAP APIs. Read literally, it has the potential to change the architecture of agentic AI projects across every SAP customer landscape. Read carefully, it is also more interesting than the lock-in headlines suggest. The policy targets a specific category of AI behavior, not AI as such. It connects to commercial mechanics that go well beyond API stability. And the literal text, in its current form, will probably not survive the next two policy revisions intact. There is a lot to unpack. I will walk through what the policy actually says, how the SAP-watching community is reading it, what the rest of the major enterprise vendors are doin...

The Illusion of Value: Why Salesforce’s Agentic Work Unit is the New "Bad Query" of the AI Era

The News On February. 25, 2026, Salesforce announced a pricing and metrics update . During the company’s Q4 FY2026 earnings call, CEO Marc Benio ff, together with CMO Patrick Stokes , unveiled the Agentic Work Unit (AWU). Positioned as a metric to quantify the labor performed by autonomous digital systems, Salesforce defines an AWU as one discrete task accomplished by an AI agent. According to Salesforce, this discrete task represents the exact moment " raw intelligence is converted into real work ". It is not a fixed unit but measured as a processed prompt, a completed reasoning chain, or an invoked tool. Salesforce explicitly designed the AWU to move the industry conversation away from the raw consumption of Large Language Model (LLM) tokens. As Benioff noted, tokens only measure "how much an AI talks," whereas the AWU is intended to measure actual business execution. The scale of this rollout is massive. Salesforce reported that its platform has already processe...

LLM Showdown: Comparing ChatGPT, Gemini, and Grok for Automated News Research

The analyst’s day is full of research. Now, this is the age of AI and AI is here to help, isn’t it? As everyone is talking about copilots and AI agents, why not using the tools at hand to do a little research on research. NB., no one really has a good definition of an AI agent, so this might become an additional topic for research. But I digress. Imagine the following project at hand, which is not only interesting for analysts, btw, but also for a variety of roles in the corporate world. Let’s call it vendor (competitor) monitoring. The job is the following: Research reputable sites for news about a number of vendors, relating to a set of keywords. Reputable sites are high quality news sites, high quality tech publications, high quality analyst sites and, of course the news pages of the vendors in question. Limit the time frame of the search matching to the cadence of my information requirement, e.g., “yesterday” for a daily update or “last week” for a weekly update. Provide a summary ...

The CDP is dead – long live the CDP!

In the past few years, I have written about CDPs, what they are and what their value is – or rather can be. My definition of a CDP that I laid out in one of my column articles on CustomerThink is:  A Customer Data Platform is a software that creates persistent, unified customer records that enable business processes that have the customers’ interests and objectives in mind. It is a good thing that CDPs evolved from its origins of being a packaged software owned by marketers, serving marketers. Having looked at CDP’s as a band aid that fixes the proliferation of data silos that emerged for a number of reasons, I have ultimately come to the conclusion and am here to say that the customer data platform as an entity is increasingly becoming irrelevant – or in the typical marketing hyperbole – dead.  Why is that? There are mainly four reasons for it.  For one, many an application has its own CDP variant already embedded as part of enabling its core functionality. Any engageme...