The development that matters most
On September 28, 2026, Meta Platforms announced a new Meta Enterprise Platform, to be led by Chirantan Desai.1 The same day, Desai, known as "CJ" and about one year into his tenure as MongoDB's CEO, announced he had left that job to head the new unit. He will not be its CEO.2 A consumer-technology giant has hired the head of a well-known business database company to sell to businesses. That is why this story belongs at the top of an enterprise-software piece.
Our tracked narrative on this theme sorts the response into three groups: big technology firms entering, startups raising money, and established software vendors defending their positions.3 Our dossier also names guidance raises at UiPath, a security acquisition by Fortinet of Virtue AI, and 82% cloud growth at Alphabet as part of that narrative.3 It holds no underlying filings or articles for those three items. We therefore report them only as part of the narrative and add no figures or detail of our own.
Startups: small cheques, narrow jobs
The money going into AI-native startups is aimed at specific, unglamorous tasks. On September 24, 2026, Dextr AI came out of hiding with $6.7 million in seed funding to build AI agents for hotel reservations, guest requests and staff coordination. Elevation Capital led the round, with Foundation Capital taking part.4 Our records list Sajid Shariff and Scott Arnold as founders. They list San Francisco as the headquarters and name two products, Alfred and Daisy.5 Crunchbase News reports that Shariff began by asking former hospitality colleagues what had changed since he left the industry. Their answer was rising labour costs, staffing gaps and pressure to bring in more revenue.6
On September 10, 2026, Latitude, a payments platform founded by Cyril Mathew, formerly of Stripe's crypto team, raised a $35 million Series A. It converts stablecoins (digital tokens pegged to a currency) into local currencies.7 Our records also show that Lightspeed Venture Partners, Wellington Management, Goldman Sachs Alternatives and Tiger Global invested in Temporal Technologies, which is based in Bellevue, Washington.5 The dossier says nothing more about that company, so we draw no conclusion from it.
Vertical specialists make a similar pitch. Dentira's CEO, Vikas Gupta, says the company provides procure-to-pay services (the buying and paying process) for healthcare groups with hundreds, sometimes thousands, of clinics.8 He also says: "Dentira is already the biggest player in a few verticals, and we're planning to become the largest vertical software provider in procurement technology overall."8 That is the company's own claim, made in an interview, and we have not checked it independently.
Where the next disruption may come: finance and back-office software
Sean Jacobsohn of Norwest, a venture investor, gave Crunchbase News the most useful view of where incumbents are exposed. He says: "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 In plain terms, the giants' main products look safe for now, while their side products do not.
He also points to who is buying. Finance chiefs choose their own software with less sign-off from others, which makes it easier to displace what they already use: "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
Light is a finance-software company that aims at this opening. Its VP of Marketing, Luke Richardson, says it serves fast-moving businesses with multiple legal entities. The company is headquartered in Europe, operates worldwide, and targets firms of 50,000 to 5,000 employees, in his words.10 The range as quoted runs from larger to smaller, so we reproduce it as stated.
One of Jacobsohn's habits is a useful test for readers judging any startup's claims. He says: "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."9 He also says he has built a "Failure Museum" of more than 1,500 items from failed companies and products.9
The limits: maths, rules and control
The case for AI in finance runs into a plain obstacle, and the same investor states it: "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 For accounting, tax and reporting, that is a serious restriction. Much of the near-term opportunity may sit in tasks where a number does not have to be exact.
The second limit is control. In a piece presented by database company EDB, which has a commercial interest in the answer, the argument is that rules must be built into the data layer. EDB writes: "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."11 Our narrative adds data sovereignty, meaning where data is stored and which country's law governs it, as a further brake in regulated finance.3
How the incumbents look from the inside
One defender is French software company 74Software. It says it serves more than 12,000 companies, including over 1,500 financial-services customers.12 Between July 20 and 24, 2026, it bought back 24,777 of its own shares at a weighted average of €35.49 each, for a total of €879,243, all on Euronext Paris.12 On July 22 alone, it bought 12,970 shares at €35.42.12 A company buying its own shares is a signal of management's view of its own price. It does not show how AI will affect its business.
A caution on source quality: our own checks show that only 31% of 2,927 claims we tested from the distribution service carrying this notice held up.12 We rely on the figures here because they come directly from the company's own disclosure. Even so, read this as a sign of reliability problems in the feed, not as a reason to doubt the notice. The Crunchbase News material, which supplies the Dextr AI and Norwest items, has a 50% measured hold-up rate across 38 claims.4 That is better, but not high. We use it chiefly for what named people said, and attribute those statements accordingly.
Elsewhere, Sprinklr CEO Rory P. Read sold 145,865 shares on September 16, 2026, at a weighted average of $5.55, according to an SEC Form 4 filing.13 The dossier does not say why, and executives sell for many reasons. We offer no interpretation.
What our evidence cannot tell you
Our dossier holds funding announcements, executive moves and interview claims. It holds no customer counts, revenue figures or usage data for any AI agent product. So we cannot say whether agents are replacing existing software or being tried alongside it. Everything above says where money and attention are going, not which side is winning.
What to watch
- What Meta's enterprise platform actually sells, and who joins Desai. The dossier says only that he will not be CEO.2
- Whether finance and back-office startups win customers away from the secondary products of the large vendors, as Jacobsohn suggests.9
- Whether vendors can show in practice that agents stay inside company rules, as EDB's argument requires.11
- Whether 74Software's buyback programme continues in later weekly disclosures.12
