Okta's rally and a16z's market data point the same way: investors pay for accelerating bookings, new products and cash flow, not headline growth alone.
Okta made new AI agent controls generally available, said a gateway, endpoint discovery and an expanded kill switch are planned for later quarters, and joined AWS, Salesforce, ServiceNow and others in an alliance to standardize agent security.
1Password's technology chief says agents should prove each task was done correctly before getting access for the next, an approach reflected in products it launched in July.
Almost every company surveyed has a written AI access policy, Delinea says, yet fewer than one in five can catch an AI tool or agent overstepping as it happens.
The AI app builder added $100 million in annualized revenue since June and now lets companies run apps it creates behind Microsoft sign-in and IT policies.
Two new products tackle different sides of the agent problem, while an Omdia survey of about 400 security leaders points to confusion over threats and mixed views on platforms.
The observability startup is buying software that adjusts CPU, memory and node scaling on its own, signaling a shift from monitoring infrastructure to acting on it.
Docker now runs AI coding agents in isolated cloud microVMs billed by the second, and is taking its agent permission format to the Cloud Native Computing Foundation.
JetBrains regrouped its AI coding products into a three-part system for individual developers, teams and company-wide oversight, with shared credits and multi-vendor agent support.