Experian ships an Agent OS with ServiceNow as its first partner
The week's enterprise-AI theme is "kill switches and minimum access." Today it came from a credit bureau.
Experian launched an "Agent OS" today, putting ServiceNow's agents at the front of the queue and selling the rest of the package — identity controls, logging, model selection, a common gateway, a Model Context Protocol server — to every other ServiceNow customer. The credit-rating giant has spent years running machine learning underneath consumer-finance products; today's news repackages that as a platform play, with chief AI officer Vijay Mehta positioning the move as "the enterprise scaling phase" rather than another PoC. ServiceNow becomes the first deployment partner, with its agents wired into Experian's Ascend analytics platform so financial-services customers can pluck credit data, model-risk checks and decisioning into existing workflows without a bespoke build.
The interesting story is the controls layer. "No agent can escape into the wilderness without us knowing and without us being able to kill it," Mehta said. Agents get minimum permissions, their activity is logged, an "auditor" agent polices the others against regulatory rules, and a semantic layer plus knowledge graph keep data definitions consistent so the platform survives when the underlying model gets swapped out. Experian can also route cheaper tasks to smaller models rather than paying for frontier inference on every call — a more candid acknowledgment of cost economics than most enterprise announcements bother to make. Early customers are insurers, lenders and financial-services providers using Experian's model risk management service; employee onboarding and verification are the second wave.
Google's Gemini Spark can now edit, curate and build albums inside Google Photos. The personal agent — Google's answer to ChatGPT-as-everything — was already wired into Workspace and Search; today it picked up the Photos library. Users connect Photos to Gemini, flip on Spark in the app, and ask the agent to turn concert flyers into calendar events, assemble shared albums of favorite shots, or run multi-step workflows across their library. It's a small feature on paper and a signal in context: Google is hunting for the consumer use case that makes a personal AI agent feel indispensable, and giving Spark the keys to the largest photo library most users own is the biggest single-product surface it's handed an agent so far. Rollout is gradual and there's no word on international expansion.
What to watch: whether Experian's "kill the agent" controls become a reference architecture other regulated enterprises copy — and whether Spark in Photos is the consumer moment that finally makes Google's personal agent stick.
Should enterprise AI agents carry the same kind of "no escape into the wilderness" guarantees Experian is rolling out? Tell us in the comments.
