The money is following control, not the agents
Salesforce said on October 20, 2026 that it had signed a definitive agreement to buy Braze for about $5.66 billion in enterprise value, paying $47.50 in cash per Braze share.1 For scale, that is more than the $3.925 billion full-year revenue midpoint Trimble kept in its own guidance this month.4 Braze is a customer engagement platform that, per the announcement, responds to customer behaviour in real time across apps, websites, email, SMS and WhatsApp.1 In other words, it is a stream of live behavioural data, which is exactly what agents need to act on.
Five days earlier, ServiceNow announced AI Passport, which it says verifies the scopes of first- and third-party AI agents in real time. It also expanded its Super Intelligence Control Tower to cover discovery, visibility, control and cost-to-value measurement on a single data model.2 Jon Sigler framed the problem this way: As AI agents proliferate across the enterprise, visibility alone isn't enough. Enterprises need confidence that every AI agent is operating within its authorized scope.
2 The company's Control Tower Alliance, part of the topic we are tracking, is built on that same Control Tower product. We have no source document in this dossier describing the Alliance's members or terms, so we do not characterise them.
Via News's analysis of these items suggests a pattern: the spending is going to the layers around agents (who an agent is, what it may touch, and what data it sees) rather than to the agents alone. That is our reading of the filings, not a claim any of these companies made.
The plumbing layer: Confluent's Streamhouse
On November 4, Confluent introduced its Streamhouse architecture and made Lightning Tables generally available on AWS. It describes this as completing a five-pillar data streaming platform: Stream, Connect, Process, Govern and Query.3 Lightning Tables is built on that platform, and the platform is built on Streamhouse.3 Zoom's addition of a context layer is part of the same topic, but this dossier holds no source document on it, so we cannot say more than that it is reported.
The common thread with Braze is the same: agents are only as useful as the live data and context they can reach. Vendors appear to be buying or building that layer before their rivals do.
The caution from the finance side
The other half of the story is about who gets displaced. Sean Jacobsohn, a Norwest partner interviewed by Crunchbase News, said disrupting the core products of Workday, ADP, SAP, UKG and Dayforce would be very hard, but it's easier to disrupt some of their secondary products, where the category isn't their core business.
5 He also made a structural point about finance software: 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.
5
He was blunt about the limits of AI here: 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.
5 That is a useful counterweight to agent hype. An investor looking to unseat incumbents is saying the opening exists, but not in the part of the product that does the arithmetic.
Note the source. The Crunchbase News interview is undated in our record, and Via News's measurement shows 49% of 41 checked claims from that outlet held up.5 These are one investor's opinions, and the small sample means the reliability figure is itself uncertain.
Agents in the wild: small but real numbers
Real deployments are visible at the small end. Dextr AI, which sells AI agents to hotels, says its agents now handle more than one million interactions a month across hundreds of hotels.6 That is a company claim, and the same Crunchbase News source carries the 49% reliability rating.6
Tiny Ltd. completed its purchase of Oso Cloud on October 6 for about $1.9 million in cash. Oso is an authorization and access-control software business with about $5.3 million in annualized recurring revenue and roughly 80 customers, all on recurring subscriptions.7 We note, as our own observation, that authorization is the same problem ServiceNow's AI Passport targets. The Tiny announcement does not say it was bought for that reason.
Results are uneven
Trimble reported third-quarter revenue of $967 million, up 7%, with record annual recurring revenue of $2.564 billion, up 10%.4 It kept its full-year revenue midpoint at $3.925 billion but lowered its recurring-revenue growth outlook to about 10% from 12% to 14%.4 It also flagged a roughly $90 million headwind to 2027 revenue and operating income growth, caused by the absence of a January 1 in fiscal 2027; recurring revenue and free cash flow are unaffected.4 The QNX royalty backlog passing $1.2 billion is part of the topic narrative, but our dossier has no primary document for it.
At the micro-cap end, TEN Holdings (Nasdaq: XHLD) reported third-quarter revenue up 17.9% to $640,000. CEO Virgilio Torres said that came from customer engagements that had been initiated in prior periods
plus new customers.8 Torres said the company cut its annual operating costs by about $2.5 million over 12 months, partly through the integration of AI-driven tools to streamline engineering and accelerate product development.
8 The quarter included a $1.8 million write-off of previously capitalised service contracts and about $0.5 million in legal costs, which the company calls non-recurring.8 Those two charges together are several times the quarter's revenue. Short-term related-party borrowings fell from $4.6 million to about $0.1 million, and total liabilities fell about 74%, from $7.7 million to $2.0 million.8 Those are the company's own figures from a press release.
A note on how much to trust the sources
Most of the press releases here come from a wire feed that Via News's checks rate at 54% of 6,584 claims holding up.1,2,8 That is a measured record, not a verdict on any one release. We have tried to attribute each claim to its speaker for that reason.
What to watch
- Whether the Salesforce: Braze deal closes on the stated terms, since the dossier holds only the signing announcement.1
- Whether Trimble's recurring revenue growth lands near its revised 10% outlook, and how the $90 million 2027 headwind plays out.4
- Whether TEN Holdings can show growth without the one-off charges, now that its borrowings have fallen.8
- Whether any vendor publishes measured error rates for agents in finance workflows, which is the claim the Norwest partner says cannot yet be trusted.5

