Home / Blog / What’s new in Gemini Enterprise Agent Platform
Tech News

What’s new in Gemini Enterprise Agent Platform

The summary is truncated, so I'll pull the Google Cloud Blog post for verified facts before drafting the paragraphs.Fetching the full Google Cloud post for…

By Dillip Chowdary • Aug 04, 2026 • Source: Google Cloud Blog

What’s new in Gemini Enterprise Agent Platform

The summary is truncated, so I'll pull the Google Cloud Blog post for verified facts before drafting the paragraphs.Fetching the full Google Cloud post for verified details.Google Cloud is opening more of Gemini Enterprise Agent Platform to general use, months after launch. In a July 30, 2026 post by Mike Clark, Director of Product Management, the company said popular capabilities from Agent Runtime to Agent Identity are now available for everyone. It also pointed to CodeMender, a managed code security agent that moves teams from passive scanning to automated remediation and aims to cut zero-day risk. The update follows customer and builder activity since launch, plus earlier materials that included 13 demos and a 20-question checklist for agentic foundations.

On the runtime side, Agent Memory Bank lets builders define structured schemas that extract and keep conversation context so agents retain user preferences, past decisions, and account history across long tasks without dropping context or inflating latency. Agent Runtime can keep agents running continuously for up to 7 days, which covers multi-day work such as a week-long sales sequence or multi-stage onboarding without constant human babysitting. For control planes, Agent Identity is a native IAM type on open standards that applies least privilege, binds access to the agent runtime to limit token theft, supports non-repudiable audit of agent actions, and manages identity lifecycle so dormant credentials do not linger. Agent Gateway is the central control point for ecosystem traffic: IAM conditions and natural language rules for access, with Model Armor inline against prompt injection, tool poisoning, and data leakage. Agent Registry is the org-wide library of agents, servers, and connections so teams reuse instead of rebuild and admins can track sprawl. Observability and evaluation share one engine: Agent Observability for end-to-end tracing of reasoning, tools, and execution; Agent Evaluation for online monitors that flag degradation and behavioral drift, with pre-built metrics, custom Python, LLM-as-a-judge, or adaptive rubrics co-developed with Google DeepMind.

Advertisement

Tech Pulse Daily

Get tomorrow's pulse first

Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.

For engineers, the practical shift is from chat demos to production agents that hold state, run for days, and leave an audit trail. Memory Bank and multi-day Runtime reduce the need for custom state stores and brittle restart logic for prospecting, supply-chain monitoring, or incident response. Identity and Gateway put agent permissions on the same footing as service accounts: least privilege, bound credentials, and a single enforcement point instead of ad hoc API keys. Registry and shared eval/observability tooling give platform teams a catalog and a single metric path from build to production, so what you grade in staging is what you watch live.

The competitive angle is platform completeness for enterprise agent fleets, not a single model drop. Google is packaging runtime stamina, memory, IAM-grade agent identity, gateway policy, registry, and production evaluation as one stack, with customer examples from AT&T (Memory Bank for sales agents that resume after gaps, including planned IVR continuity), Best Buy (Agent Identity for ownership and least privilege versus orphaned service accounts), Commerzbank AG (Registry and Gateway for discoverability, policy, and audit), Liberty Global (Gateway and Registry for multi-entity governance), and WellSky (Registry for cataloging, versioning, and only vetted agents in healthcare production). Broadcom, Palo Alto Networks, and Ping Identity are called out as using Agent Gateway at scale.

If you already build on Agent Platform, the near-term work is to map long-running jobs onto Runtime and Memory Bank schemas, put new agents behind Agent Identity and Gateway before they hit production credentials, and register them so reuse and sprawl controls are real. Watch how CodeMender lands next to this stack for automated fix loops, how online evaluation monitors behave under real drift, and whether Registry becomes the required gate for production promotion in regulated shops like healthcare and banking.

Advertisement

🔎 More interesting news

5-min tech signal

Weekday briefing for engineers who skip the noise.

No spam · Unsubscribe anytime

Advertisement

✈️ CareerPilot

Your AI job-search copilot

Match your resume against live Ashby, Greenhouse & Lever openings — fit scores, job-specific resume optimization and email alerts.

Find matching jobs →

Free Tools

Browse all tools →