At Glean, we’ve been shipping agentic AI for years—our customers are prolific users of apps, agents, assistants, and chatbots in Glean today, and in fact our platform has delivered more than 50 million agentic actions over the past year alone.
Now that customers are evolving from individuals building personal agents to departments deploying agents across the business, they’re increasingly thinking about scale—how to 10x, 100x, or even 1000x their agent deployments. Deploying agents at scale not only requires robust enterprise security and governance, but also capabilities for everyone to build, use, and manage agents with ease.
Today we’re excited to announce Glean Agents, a horizontal environment for building, deploying, orchestrating, and governing agents, enabling organizations to scale AI adoption across their businesses. With Glean Agents, everyone can easily build and use agents using natural language, as well as discover and share pre-built agents in the agent library. Agent orchestration enables agents to be triggered in reaction to data or app state changes, or in response to seeing questions they can answer. Agents are powered by Glean's LLM-agnostic agentic reasoning engine, which executes complex, multi-step workflows by retrieving data from many sources— across company data and internet data— and using actions to complete work.
With Glean Agents, users can build both personal agents for individual use and business agents that reshape how organizations work. Engineering teams can use agents to learn new areas of the code base, review pull requests for accuracy and style, and streamline postmortems. Sales teams can personalize prospecting, receive call coaching, update Salesforce, and ensure legal redlines meet internal requirements. Every department embraces agents to work differently and that starts with individuals embracing agents to change the way they work.
In this blog, we’ll further detail the capabilities of Glean Agents and how everyone in the enterprise can use it to put AI agents to work across their business.
Enabling everyone to build agents, for themselves and their teams
We’ve noticed that the most difficult process of integrating AI into workflows is building agents themselves. Converting a set of instructions into something comprehensible and implementable by an agent is a complex and time-consuming task, as is collecting, managing, and securing the appropriate data and application connections to enable the agent to be productive. A dedicated team of engineers is also often required for development, making it difficult for non-technical teams to experiment with agents.
With Glean’s agentic reasoning engine and agent builder, agents become something anyone can leverage. Users can express what they want an agent to do in natural language, and Glean takes care of the rest. Under the hood, Glean's agentic reasoning engine has access to each enterprise's unique processes, data, and actions to dynamically inform the workflow, and since Glean Search has already indexed the full breadth of enterprise knowledge, the agent automatically has secure access to the most relevant data to complete its tasks.
Agent builder’s flexibility enables everyone to capture the finer details of their complex daily workflows. Glean's agent builder expands beyond single-track workflows to intelligent branching paths based on real-time decisions. With agentic looping, agents iterate on a task multiple times until certain conditions are met, allowing it to self-correct dynamically.
Glean also makes it easy to get started from square one—pre-built agent templates applicable to a variety of general tasks are automatically provided. These templates, which can be customized and built upon, can serve as a foundation for more specialized agents.
All these capabilities are built upon Glean’s Work AI platform, enabling agents to pull from company and world knowledge while retaining full permission rules from underlying source systems. Agents can only see and work with the information and actions that users have access to, and we give companies control to further narrow down this space for agents. To ensure company data is never shared externally, we keep humans in the loop to monitor agents’ creation, output and actions.
Putting agents to work
For agents to be truly useful in the enterprise, they must take a responsive and proactive role within both employee and business processes. This makes agents what they are—rather than following a set of pre-configured rules, they are capable of responding dynamically, infusing reasoning and tools to achieve a specific goal. Agents must be capable of responding not just at the current speed of your business, but also predicting what you need to accelerate.
For example, when a new transcript appears in Gong, a call-coaching agent can be made to trigger, automatically providing the user with summaries, top concerns, and next steps relevant to the user’s workflow. Glean Agents can also be scheduled, giving you daily summaries and catching you up on on company initiatives and industry trends with your morning coffee. In addition, Glean will suggest agents to users based on the company knowledge graph and queries they’ve provided to Glean, making it easier than ever to discover and apply agents relevant to their workflows.
Agent orchestration will be an ongoing journey—we will continue to evolve Glean Agents to support integration with other third-party agents and more at the frontier of GenAI.
The capabilities of agentic reasoning
The brain behind Glean Agents is the agentic reasoning engine. It can interpret natural language queries and build them out into dynamic, multi-step workflows customized for every enterprise. Glean Agents can accomplish a wide range of tasks, from finding and analyzing relevant enterprise and world data, to executing a suite of actions that can complete everyday work.
Everyone can use our agentic reasoning engine on the fly to execute more detailed queries. In Glean Assistant, simply press the deep reasoning button, and Glean will bring advanced reasoning, reflection, and execution to the task. Whether consolidating world and work data into a single response, doing complex data analysis, or completing multi-step workflows, Glean will determine and execute the steps needed to drive your work forward.
Under the hood, Glean’s agentic reasoning engine gathers background context to fully understand the goal, reflect on the scenario at hand, and build and execute against dynamic multi-step plans using a set of tools. Central to Glean’s tools are search, using enterprise data to personalize the context for every enterprise and scenario. Glean is LLM agnostic and will continue to grow in capabilities as LLM reasoning improves.
Creating an ecosystem of agents
Of course, Glean is only one part of the future of agents, and we are committed to working together with partners across the AI ecosystem to drive their development and adoption across the enterprise. Glean allows you to index and secure your data once, and share it across all of your agents, whether they’re built with Glean or another platform. Over time, Glean Agent’s orchestration capability will expand to enable you to both use personal and departmental agents in Glean and within our larger ecosystem of third-party applications.
With our APIs, you can bring Glean Agents to any application and any surface across your enterprise. In addition, you can connect data indexed by Glean with agents built anywhere, including Google Vertex AI, Amazon Bedrock, NVIDIA NIMs, Databricks, and more. We have already started agent integrations with our technical partners, including Cisco Webex, Miro, and Sigma Computing.
Empowering every enterprise with agents
Glean Agents enable everyday users and developers to embrace agentic AI and stay confidently ahead in this new era. Features like the agent library, an explorable, open-source repository of completed agents made by coworkers, make sharing agents easier than ever. This ensures that as individual trailblazers in each company make great strides with AI, those gains can be seen and shared with the entire organization.
This discoverability of agentic applications, through both Glean’s in-app agent suggestion feature and the agent library, is the foundation of our vision for the future of agents in the workplace. It enables the productivity benefits of agents, otherwise secluded in silos or held back by limited enablement resources, to instead spread to everyone—regardless of their experience and expertise in working with AI systems.
Want to discover more about how Glean Agents help us deliver the market’s most comprehensive work AI platform? Watch the Glean Live: Winter ‘25 launch event! If you’re looking for a way to transform the way you work with enterprise-ready Work AI and agents, sign up for a demo today.
Our agentic reasoning engine and deep reasoning capabilities for Assistant are in beta. Access to the natural language agent builder, unified agent and prompt library, and agent orchestration features will come soon.
