Glean Wants Enterprises to Stop Winging It With AI Agents
The Work AI company is giving CIOs a repeatable framework to build, govern, and measure agents—before the sprawl gets worse.
Enterprises have spent the past year building AI agents. Now they have to figure out what to do with them. To help, Glean has released its Enterprise Agent Development Lifecycle (ADLC), a framework that gives Chief Information Officers and IT leaders a repeatable path to scale agents across their organizations.
“Agents are software. They need a disciplined way to be defined, built, launched, governed, and improved over time,” Emrecan Dogan, Glean’s chief product officer, said in a statement. The ADLC delivers that, giving CIOs a structured approach for ensuring agents are useful, secure, and tied to measurable business outcomes.
The framework draws from Glean’s own experience. After releasing autonomous agents last December, the company found itself facing the same problem it now wants to help customers solve: thousands of agents built across teams, with no consistent way to govern or measure them. “It’s really important for us to define what success looks like, what do we want AI agents to be able to do, and create a product that allows our customers, as well as ourselves, to be able to operationalize the agents that are built,” said Selene Kim, a Glean product manager, in an interview with The AI Economy.
Glean’s Enterprise Agent Development Lifecycle features seven stages: Opportunity, Design, Performance, Input, Develop, Launch, and Monitor and Improve. “Just like the software development lifecycle, you need every single step of the flow,” Kim pointed out. It’s intentionally platform-agnostic—Kim described it as Glean’s “opinion piece” for the broader AI community, something any organization can adopt.
But Kim says the harder problem isn’t building—it’s what comes after. “The problem that we’re solving with the Agent Development Lifecycle is not just about building the agents,” she explained. “How do we make sure that those agents that are built…can you get it to work in production? How do you measure whether it is successful? How do you know when to say it’s not working as expected?”
That said, Glean acknowledged the framework is freely adoptable—but argues its advantage lies not only in its context engine but also in the platform capabilities it has built to execute every stage of the ADLC—all of which competitors would take time to replicate.
Alongside its ADLC announcement, Glean is introducing eight capabilities that address parts of the framework where most enterprises are stuck: building agents with the right context, launching them with the right governance, and measuring the value they generate over time. “They’re mapped to each part of the Agent Development Lifecycle because we know that there are many things that need to be done to have a perfect agent development life cycle,” Kim shared.
Here’s a look at the new tools and which phase they fall under:
Auto Mode Agent Builder (Develop): Using natural language, this editor generates a Glean-powered agent capable of planning, reasoning, and executing across the enterprise graph without the need for pre-defined workflows or manual configuration.
Debug and Trace Views (Develop): This provides full visibility into every action taken by the AI agent, cataloging inputs, tool calls, LLM decisions, and outputs. No longer must developers guess why their agents have failed.
Sub-Agents (Develop): These are specialized agents focused on discrete tasks, coordinated by a parent agent at runtime.
Expanded Agent Sandbox: Glean customers now have access to a secure file system and code execution within their virtual private cloud. In addition, it supports adding both apps and individual actions.
Content and Scheduled Triggers (Context): This empowers agents to automatically react to content changes, scheduled runs, forms, and external events, enabling them to operate more within existing business processes.
New Agent Library Controls (Launch): Glean has added verification badges, featured agents, departmental categories, and a soft-delete with an admin restore—all aimed at turning this marketplace into the “governed front door for agent distribution.”
Agent Access Policies (Launch): AI agents now have consistent controls, such as blocking or flagging sensitive material, thanks to organization-wide guardrails.
Updated Agent Insights Dashboard (Monitor and Improve): Glean has rebuilt this monitoring tool so it better tracks adoption, top use cases, estimated hours saved, and feedback trends over time. In doing so, CIOs will better understand which agents are delivering value, and which ones aren’t.
Glean revealed that Auto Mode Agent Builder, Debug and Trace Views, the Sub-Agents, Agent Sandbox, and the new Agent Library Controls are all generally available today. However, the Content and Scheduled Triggers and Agent Access Policies are in beta, and the Agent Insights Dashboard is coming soon.