The cheapest number in the agent race
The platform announcements get the attention. The most concrete price in our material is for a smaller piece: the layer that decides what an agent is allowed to do. On October 6, 2026, Tiny Ltd. (TSX: TINY), a holding company, said it had completed the acquisition of the assets of Oso Cloud, an enterprise software business specializing in authorization and access control, for approximately $1.9 million in cash.1 Oso brings approximately $5.3 million in annualized recurring revenue (ARR), the yearly value of its subscription contracts.1
Our arithmetic, not a figure Tiny published: $1.9 million against $5.3 million of recurring revenue is roughly 0.36 times annual subscription revenue. Put plainly, the buyer paid about a third of one year's subscription income. We have no figures on Oso's profit, costs or growth, so this is a headline ratio and not a verdict on whether the price was cheap.
The deal is small, but it lifts Tiny's own recurring revenue by about 7.6%, from $70.0 million to approximately $75.4 million on a pro forma basis, meaning as if the deal had been in place all along.1 Oso served approximately 80 customers in security, fintech and developer software, all on recurring subscription contracts including several multi-year agreements.1 Contracts of that kind are what a buyer is paying for.
One caution on reliability. Our checks found that 55% of 4,960 claims from the wire service that carried this release held up.1 The release is the company's own announcement of its own deal, so treat the figures as Tiny's statements, not independently audited numbers.
Why authorization is suddenly worth owning
The reason a business like Oso matters is a plain engineering argument. In an article presented by EDB and published by VentureBeat, the company says that as enterprises give agents the ability to plan, decide and act across systems without a human approving each step, the hard question is what actually stops an agent from completing an action it was never authorized to do.2 EDB's answer is blunt: "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."2
Note that EDB sells data infrastructure, and the piece was labelled as presented by EDB, so this is a vendor's argument for its own category. It is still the logic that explains why access-control software becomes a purchase target once agents start acting on their own.
The topic brief behind this feature also lists Harvey acquiring Guardrails AI, and Latitude raising a seed round and Dextr AI a Series A. The dossier gives no price or terms for the Harvey deal, so we make no comparison with Tiny's. We say only that these are reported as part of the same pattern.
The platforms: announced, not yet priced
The big vendors are building the layer above. At SAP Connect on October 6, SAP demonstrated what it calls the Autonomous Enterprise, with Joule Work and Joule Assistants generating value across every line of business on the SAP Business AI Platform.3 The same day, PwC and SAP said they were extending their collaboration into new areas of joint AI innovation, building on the Billing Assistant.4
Two days later, at Zeta Live 2026, Zeta Global announced AthenaOS, an enterprise intelligence operating system built on the Zeta Data Cloud.5 It also announced Athena MCP, a headless product that brings Zeta intelligence into the applications, agents and workflows where teams already work.6 Zeta added the Athena Inference Model (AIM), developed with Fireworks using NVIDIA Nemotron,7 and said it had begun work with partners on a specialized AIM chip, which it describes as a capital-light approach.8
For an investor, the useful observation is what these items do not contain. The dossier holds no revenue, pricing or customer figures for any of these launches. They are announcements of direction. If you own SAP through a broad index fund, which our material does not say, nothing here gives you a number to put against it yet. The only hard price in this story is the one on the small authorization deal.
The office of the CFO: big opportunity, hard limit
The most direct finance commentary comes from Sean Jacobsohn of Norwest, who worked at HR software companies before he began investing in them.9 He draws a firm line on what AI should do in finance: "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."9
That echoes EDB. Both say that where the numbers must be right, the system around the model has to guarantee it. The model cannot be trusted to guarantee it itself. Together they describe a product shape: agents handle the workflow, and deterministic software handles the arithmetic and the permissions.
On where the money might be, Jacobsohn is more cautious than the headline narrative. "I think it would be very hard to disrupt the core products of Workday, ADP, SAP, UKG and Dayforce. But it's easier to disrupt some of their secondary products, where the category isn't their core business."9 He adds a point about buyers: "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."9 His example is CFOs buying software for themselves.
Read together, a startup selling into the finance team has a real route in. Taking on the giants' core products head-on is, in his view, the harder path. That sits awkwardly beside platform launches from SAP, one of the five names he lists.
What the early funding looks like
At the seed end, Crunchbase News reported that Dextr AI raised $6.7 million for hotel AI agents.10 Our entity records list Sajid Shariff and Scott Arnold as founders, with Elevation Capital and Foundation Capital among its investors.10 That is a vertical agent business, not a governance tool, and it sits far from the CFO's office. It shows money going to narrow agents that do one job.
Daniel Docter of Dell Technologies Capital offers a lens for judging timing: "It's important to have a sense of whether a company is doing category disruption or category creation... in cases of category disruption, I think there's value in first-mover advantage."11 Agents in finance look like disruption of an existing category more than creation of a new one, so by his logic speed matters.
A reliability note applies to the interview pieces from Crunchbase News: 50% of 38 claims we checked from that outlet held up.9,10,11 We quote named speakers verbatim, but readers should weigh the surrounding reporting accordingly.
What to watch
- Disclosure. Whether SAP or Zeta put revenue, pricing or customer counts to the agent products. Until then they are announcements only.
- More authorization and guardrail deals. Oso's price is one data point. A second one at a similar ratio would suggest a pattern. A very different price would suggest Tiny's deal was an outlier.
- Retention at Oso under Tiny. Its roughly 80 subscription customers are the asset. Tiny's later reporting on recurring revenue will show whether they stayed.
- Where precision is required. Watch whether finance products keep calculation outside the model, as Jacobsohn and EDB both argue they should.
