Intapp's OpenAI plug-in validates the AI thesis — now AI bookings have to carry the multiple

On September 10, Intapp put its DealCloud with Celeste plug-in inside OpenAI's ChatGPT for Financial Services, letting deal teams at banks and investment firms query their own client, deal, and relationship records from inside the chatbot. It reads like a splashy AI announcement, and the press release gives it the full weight: roughly 1,700 Intapp clients in private capital and investment banking, compliance rules "enforced regardless of where the knowledge is accessed," and a flagship OpenAI model providing the reasoning.

What the release does not include is any number. No bookings, no revenue attached, no ARR contribution. That gap matters, because it tells you exactly how to read this announcement: it is a distribution and validation event for Intapp's AI bet, not proof that the bet is paying off. The evidence for whether that bet works lives in last quarter's earnings, and it is more convincing than the press release on its own would suggest.

Who captures the value in the partnership

The reason this collaboration is strategically coherent rather than a warning sign is the division of labor. OpenAI brings the general reasoning engine and the branded "ChatGPT for Financial Services" experience. Intapp brings the governed data — a firm's clients, funds, investors, and the judgment calls partners have made over years — plus the compliance layer that keeps material non-public information behind information barriers. That is precisely Intapp's own sales pitch on its earnings call: frontier models supply generic knowledge, while Celeste works from firm-specific data and compliance rules that a generic model cannot legally or practically reproduce.

Putting Intapp's governed records inside ChatGPT is not Intapp handing its customer relationship to OpenAI. It is OpenAI renting Intapp's moat to reach regulated financial clients it could not serve on its own. For Intapp, that is cheap distribution into a product where its target buyers are already working. Thad Jampol, Intapp's co-founder and chief product officer, framed it as the firm-specific knowledge that "cannot be bought". The collaboration reinforces the thesis; it does not weaken it.

The operating proof the stock actually turns on

The announcement is a footnote to the debate that has actually moved the stock. Intapp's shares fell sharply earlier this year, recovered roughly 50% over the past four months, and are down about 8% over the past week and roughly 17% year to date. That recent erosion is worth putting next to the underlying numbers from the June quarter.

SaaS revenue grew 27% year over year to $115.0 million. Cloud annual recurring revenue reached $495.7 million, up 29%, and now represents 84% of total ARR. Trailing-twelve-month cloud net revenue retention held at 123% — meaning existing clients are spending roughly a quarter more each year without being replaced. Free cash flow for fiscal 2026 was $144.7 million, a 25% margin, which management said had reached the range it had targeted for fiscal 2029. Non-GAAP gross margin ticked up to 79.4% from 78.0% a year earlier.

Those are the numbers that matter. The OpenAI deal gets the AI story distribution; the retained margins and 29% ARR growth say the story has a durable engine behind it. On the question that should decide how you read the stock — is the AI narrative funded by real demand or is it a pitch — the June quarter gave a concrete answer: AI bookings were more than 20% of net new bookings, roughly double the prior quarter, and Celeste reached general availability on July 15 with a pricing model of a platform fee plus a usage fee.

Why the multiple needs proof

Set that against the price. At about $38, Intapp trades at roughly 24 times the fiscal 2027 non-GAAP EPS guidance of $1.58 to $1.62, and about five times trailing sales. That is not expensive for a software company growing cloud ARR 29% with 123% retention and a 25% free-cash-flow margin. The reset during 2026 did the work: the multiple has absorbed a lot of bad news, and the risk-return has rebalanced toward the operating record rather than the narrative.

The strongest bear case is that the AI monetization is still early and unproven, and the transition story carries real friction. License revenue fell 25% year over year as on-premise clients moved toward cloud, and remaining performance obligations grew only 16% because on-premise deals were a drag. That means reported growth is being slowed by the deliberate wind-down of an old business, and the offset — paid AI — has only a few months of evidence behind it. A platform-plus-usage model for Celeste could scale, but "could" is not a quarter yet.

The honest read

The OpenAI collaboration is real support for the AI thesis and a modest distribution win, but on its own it changes nothing about the case. The case is that Intapp has durable, high-retention cloud growth, expanding margins, and early signs that AI attaches real revenue on top of it — and that the shares, after the reset, are priced more like proof is still needed than as if it had already arrived.

The proof window is the next two to four quarters: whether AI bookings continue to grow as a share of new business and whether cloud ARR holds near 29% as the on-premise drag rolls off. If AI keeps compounding and ARR holds, a roughly 24-times-forward multiple on a 25%-free-cash-flow business starts to look thin. If the AI attach rate stalls, the multiple has little room to be wrong. The evidence today justifies watching closely and leaning constructively — but the collaboration is a reason to keep that faith, not a number to buy yet.