What the evidence supports, and what it doesn't
European business readers will want to know what the agent-platform race means for EU companies and EU regulation. The honest answer is that our verified dossier says very little about either. It contains no EU rule, regulator or European buyer reacting to these launches. It does contain one European-headquartered finance vendor, Light, plus the platform launches and a pointed argument about who enforces the rules when software acts on its own. We lead with that, and we mark where the evidence stops.
The platform giants are announcing, not yet proving
At SAP Connect on 6 October 2026, SAP demonstrated what it calls the Autonomous Enterprise. It showed Joule Work and Joule Assistants generating value across every line of business, built on the SAP Business AI Platform.1 The same day, PwC and SAP said they would extend their collaboration into new areas of joint AI innovation, building on the Billing Assistant.2 The dossier gives no customer numbers, savings figures or error rates for either announcement. These are vendor demonstrations, and we treat them as such.
Two days later, at Zeta Live 2026, Zeta Global announced AthenaOS. It describes this as an enterprise intelligence operating system built on the Zeta Data Cloud that builds dynamically around an enterprise team's objective.3 It also announced Athena MCP, a headless product meant to bring Zeta intelligence into the applications, agents and workflows teams already use.4 A third announcement, the Athena Inference Model, was developed with Fireworks using NVIDIA Nemotron and targets specific AthenaOS use cases.5 Zeta also said it has begun work with partners on a specialised AIM chip, which it describes as a capital-light approach.6 Our topic summary also lists Meta's Enterprise Platform, with MongoDB's CEO moving over, as part of the same race.7 The dossier holds no source document for Meta's move, so we report it only as part of that summary.
The real question: who stops an agent?
The sharpest material in the dossier concerns governance rather than capability. In a piece presented by EDB and published by VentureBeat AI, the vendor frames the question every architecture review now faces. When an agent attempts something it was never authorised to do, what actually stops it?8 Its answer: "The enterprise should not rely on a model choosing to follow policy. The policy has to be enforced by the system. That is the difference between hoping an actor stays in bounds and constructing bounds it cannot cross to begin with." (EDB)8
Readers should note that this is a sponsored, vendor-presented piece. EDB sells data infrastructure, and the argument supports its own product category. The logic is still worth understanding. A rule that exists only as an instruction to the model can be ignored by the model. A rule built into the data layer cannot. For European firms already accustomed to compliance obligations, this is the version of the argument most likely to matter. We repeat that our dossier cites no specific EU regulation.
Where the money is going: the supporting layers
The smaller deals suggest investors and acquirers agree that control is a product in its own right. Tiny Ltd. completed its acquisition of Oso Cloud, an authorization and access-control business, for approximately $1.9 million in cash.9 Oso brings about $5.3 million in annualised recurring revenue and lifts Tiny's total from $70.0 million to roughly $75.4 million, an increase of about 7.6%.9 Tiny says Oso served approximately 80 customers in security, fintech and developer software, all on recurring subscription contracts, several of them multi-year.9 The price is small. This is a modest software business changing hands, not a bet on the whole market. The press release is also Tiny's own account, and the source's measured reliability is mixed: 55% of 4,960 checked claims from this outlet held up.9
Our topic summary adds that Harvey acquired Guardrails AI and that Latitude and Dextr AI raised seed and Series A rounds.7 We can add detail only for Dextr AI. Crunchbase News reports that it raised $6.7 million for hotel AI agents,10 and our entity records list Sajid Shariff and Scott Arnold as founders, with Elevation Capital and Foundation Capital as investors.10 Crunchbase News is a weaker source than it looks. Only 50% of 38 checked claims from it held up, so the funding figure deserves independent confirmation.10
Finance: promising target, with a catch
The topic's finance angle comes from Norwest's Sean Jacobsohn. He says finance is among the most promising areas to disrupt, and one reason is buying behaviour: "There's actually one less layer of approval when they're buying their own software, so it is a little easier to replace it when they're the direct buyer."11 He is also blunt about limits: "You don't want AI doing calculations because it is not good at math. There are certain workflows it can handle where it doesn't produce precise numbers. But when you need precision, accuracy and calculations, you can't rely on AI for that."11
This sits awkwardly beside the autonomous-enterprise pitch. If precision work cannot be trusted to the model, then the governance layers above, which keep agents away from the books, matter more. Jacobsohn also argues it would be very hard to disrupt the core products of Workday, ADP, SAP, UKG and Dayforce, while their secondary products are easier targets.11 That fits SAP's strategy of building agents on top of its own platform rather than leaving the field open.
The one European data point
Luke Richardson, VP Marketing at Light, says the company is headquartered in Europe but operates worldwide, serving fast-moving businesses with multiple entities.12 That is the only European company in our dossier. It shows that a European finance-software vendor is competing for exactly the multi-entity customers these agent platforms are chasing. It tells us nothing about how Light's products perform or how it compares with SAP.
What to watch
- Evidence over demonstrations. Look for SAP or PwC publishing customer results from the Billing Assistant work, not just showcase announcements.2
- Whether governance becomes a purchase requirement. The Oso deal, at about 80 customers, is a small signal. A larger acquisition in access control would be a stronger one.9
- How Zeta's AIM chip effort develops. The company describes it as capital-light, and it is at an early stage.6
- European regulators and buyers. If they begin to specify system-enforced controls on autonomous agents, the EDB argument becomes more than a vendor position. Nothing in our current material shows that yet.
