The biggest enterprise software companies are rebuilding themselves around AI agents, which are programs that act on a company's behalf. They are doing it in two ways at once. They are buying up the data that feeds those agents, and they are building tools to keep the agents on a leash. The verified record also shows that the results are uneven. A profitable mid-sized vendor has trimmed its growth outlook, and a micro-cap company is still working through write-offs.
Consolidation: Salesforce pays $5.66 billion for Braze
On October 20, 2026, Salesforce (NYSE: CRM) said it had signed a definitive agreement to buy Braze (NASDAQ: BRZE) for approximately $5.66 billion in enterprise value, with Braze Class A shareholders receiving $47.50 in cash per share.1 Braze is a customer engagement platform. It helps marketers respond to customer behaviour in real time across mobile apps, websites, email, SMS and WhatsApp.1 The logic is the one running through this whole story. An AI agent is only as useful as the live customer data it can see, so owning that stream is worth paying for.
Smaller deals follow the same pattern. Tiny Ltd. (TSX: TINY) completed its acquisition of Oso Cloud for approximately $1.9 million in cash.7 Oso is an enterprise software business specialising in authorization and access control. It adds approximately $5.3 million in annualized recurring revenue, lifting Tiny's total by about 7.6%, from $70.0 million to roughly $75.4 million on a pro forma basis.7 Oso served approximately 80 customers in security, fintech and developer software, all on recurring subscription contracts, including several multi-year agreements.7
Governance: who is allowed to do what?
As agents multiply, the next problem is controlling them. ServiceNow (NYSE: NOW) announced a new AI Passport solution on October 15, 2026. It verifies the scopes of first- and third-party AI agents in real time, and it sits inside what ServiceNow calls its Super Intelligence Control Tower.3 Jon Sigler framed the gap in these words: "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."3
ServiceNow's separate Control Tower Alliance is built on the same Control Tower product, according to our entity records.3 Confluent made a parallel move on the data side on November 4, 2026. It introduced its Streamhouse architecture and made Lightning Tables generally available on AWS. That completes a five-pillar data streaming platform covering Stream, Connect, Process, Govern and Query.6 Both announcements are vendor statements about their own products. Neither is independent evidence that customers are getting value from them.
The finance stack: an opening, with a warning attached
Venture investors see room to attack older business software. Sean Jacobsohn, a partner at Norwest, argues that disruption is far easier at the edges than at the core: "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."4
He also points to a structural opening in finance. "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," he says of chief financial officers.4 That is the opening for challengers to older finance tools.
His warning matters just as much for this audience: "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."4 This is one investor's view. It is still a useful counterweight to the idea that agents will simply take over the books.
Jacobsohn also describes how he judges young companies. "Before I invest, I'll go on a lot of sales calls I set up with the CEO to see how good they are at selling," he says.4 He has built a "Failure Museum" of more than 1,500 items from failed companies and products, which he studies to understand why they failed.4
Uneven results: Trimble and TEN Holdings
Trimble's third quarter looked solid. Revenue was $967 million, up 7% from a year earlier. Earnings per share were $0.86, and annual recurring revenue, the money from subscriptions that renew, hit a record $2.564 billion, up 10%.5 Gross margin was 70.5%.5
The outlook was more guarded. Trimble kept its full-year revenue midpoint at $3.925 billion and nudged its earnings-per-share midpoint up to $3.66. It cut its expected growth in annual recurring revenue to about 10%, from a previous range of 12% to 14%.5 For 2027, the absence of a January 1 in the fiscal calendar creates an approximately $90 million headwind to revenue and operating income growth. Trimble says annual recurring revenue and free cash flow are unaffected.5 That is a calendar quirk, not a collapse in demand. The lowered subscription-growth forecast is the more telling detail.
At the small end, TEN Holdings (Nasdaq: XHLD) reported third-quarter revenue up 17.9% to $640,000. Chief Executive Virgilio Torres attributed it to completing "customer engagements that had been initiated in prior periods" and adding revenue from new customers.4 The quarter also carried two expenses the company calls non-recurring: a $1.8 million write-off of previously capitalized service contracts and approximately $0.5 million in legal expenses.4 Put plainly, those one-off charges are several times larger than the quarter's revenue.
The balance sheet did improve. Short-term related-party borrowings fell from $4.6 million to about $0.1 million, and total liabilities fell roughly 74%, from $7.7 million to $2.0 million.4 The company says it has cut its annual operating cost base by about $2.5 million, partly through AI-driven tools.4 That is an example of agents cutting costs in a real company, though a tiny one.
What the evidence cannot yet tell us
The AI-agent economy is also visible in startup funding. Dextr AI, which builds agents for hotels, says its agents now handle more than 1 million interactions a month across hundreds of hotels.8 That figure comes from the company itself.
We also have no source documents for some parts of the original story. These include Zoom's context layer and the reported figure of QNX's royalty backlog passing $1.2 billion. We have not verified them, so we do not rely on them here.
Source quality matters, too. The press-release wire documents behind much of this piece held up in 54% of 6,584 claims we checked. The Crunchbase News interviews held up in 49% of 41 claims checked.1,4,8 Both are well below a standard we would call settled. That is why we lean on direct quotes and attribute every company claim to the company that made it.
What to watch
- Whether the Salesforce–Braze deal closes on the stated terms. The agreement is signed, but it is not yet a completed deal.1
- Trimble's fourth quarter. It has guided to revenue of about $1.046 billion, up 8%, and earnings per share of $1.16.5 Watch whether annual recurring revenue growth holds near 10%.
- Whether TEN Holdings' revenue grows without further write-offs once the restructuring charges stop.4
- Whether vendors publish usage or customer-outcome figures for their governance tools. The announcements so far describe capabilities, not results.3
