Okta's partners touched all 20 of its biggest deals. Agent identity still has no number.

Take the consensus figure out of the print and the beat is $11.97 million, roughly 1.5 percent ^[1]. Growth of 11 percent on $805 million implies a year-ago quarter near $725 million and a twelve-month gain of about $80 million ^[3][2], which makes the entire beat around 15 percent of the revenue Okta added over a year ^[3]. Whatever the AI pipeline converted in the three months to July 31 ^[2] sits somewhere inside that, and the interview puts no number on it.

That matters for how the channel reads the signal. Partner involvement in each of the 20 largest deals ^[1] is a statement about the top of the deal distribution, not an attach rate across the book, and the largest deals are precisely the ones where an integrator is structurally hard to avoid. The one concrete example Eric Kelleher gave CRN was a multi-technology implementation at a large enterprise that included AI agent security ^[5]. One deal, unnamed customer. The product it would have been built on, Okta for AI Agents, has only been generally available since April ^[7], so any bookings from it are two quarters old at most.

The demand mechanism Kelleher describes is more durable than the quarter. He says agents typically inherit their privileges from the human user they were built for, which is forcing customers to look again at the human side of their identity estate ^[8]. That is a spend argument that does not depend on anyone buying a new agent SKU.

The independence pitch is worth separating from the vendor's own claims because it comes from a partner, not from Okta. Lucas Marquardt of Alchemy Technology Group told CRN that Okta's position as a standalone identity platform, rather than a vendor selling both applications and the identity layer beneath them, becomes more important at the agentic layer ^[11]. Alchemy has won Okta Partner of the Year six times and sits at No. 142 on CRN's 2026 Solution Provider 500 ^[10], so it is an interested witness on a vendor whose product line it sells. His underlying point, that zero trust cannot be built without first establishing who or what is asking ^[9], is the reason identity gets budget before detection does.

Okta is filling the rest of the surface by purchase as well as by build: CRN reports it is acquiring Permiso Security to extend identity threat detection ^[13], alongside its own moves into identity governance and privileged access ^[12]. That is a company assembling a category faster than it is reporting on one. Until agent identity appears as a disclosed line, the case that enterprises have opened a new budget for it rests on a pipeline comment ^[6] and a single large implementation ^[5].