- Glean is introducing independent agents: AI coworkers built on its context layer that work proactively across enterprise tools, not just in a single channel.
- These agents combine identity with context, memory, proactivity, and accountability so they can act independently while staying secure, governed, and traceable.
- The first example, OnCall Assistant in Slack, helps engineers investigate incidents, surface context, draft fixes, and accelerate response while improving over time.
Six years ago, Glean made an early bet on a different way to build enterprise AI by starting with the realities of work and building in a way that truly serves customers. We believed work has always been multiplayer and that AI should show up where people already work. People shouldn’t be forced into new systems or behaviors, or asked to stitch together solutions channel by channel and tool by tool. Instead, by integrating a secure shared view of company knowledge through permissions, indexing, and a knowledge graph, AI can act as a true multiplayer coworker that brings the right context into workflows without tedious setup or delays caused by needing to be tagged into conversations first.
Those beliefs drove Glean’s product decisions from the beginning. Glean was built to bring AI into the flow of work, grounded in enterprise context, security, governance, and the systems teams depend on every day. We innovated by building solutions that fit the way organizations actually operate. It is gratifying to see more of the market moving in that direction. Announcements like Anthropic’s Claude Tag are strong validation of an approach Glean believed in early and has continued to invest in over time.
Today, we are continuing with that same pace of innovation by introducing independent agents—AI coworkers built on top of Glean’s context layer that work alongside every team to proactively manage and execute work. They infer what needs to happen based on their memory and context, and work across surfaces—not just in Slack—so you can engage the same agent in enterprise apps like Jira, Slack, and Teams and have context carry seamlessly across them.
Independent agents who act as your teams’ AI coworkers
Enterprises face a critical trade-off when implementing agents: how do you unlock productivity by letting agents act independently alongside your employees, without breaking critical systems?
That's why we set out earlier this year to create the best of both worlds by building a new class of agents – independent agents. These agents enable autonomous execution across the enterprise without requiring teams to accept open-ended risk. Glean grounds each agent in your company’s context, processes, permissions, and connected applications, so it can understand the work, investigate issues, and take the next best step. Our security and governance capabilities ensure that everything an agent does is traceable, and remains within the guardrails set by your organization.
Independent agents in Glean have four key characteristics: identity with context, memory, proactivity, and accountability. Each characteristic enables agents like our OnCall Assistant, which partners with engineers during escalations, to use their freedom to act in line with your company’s needs.
Identity with context
Each independent agent operates with its own identity, with specifically provisioned access to the resources it needs, rather than being limited by any one user’s permissions. That way, it's possible to control what the agent can and can't see, and trace what actions it has taken across your company's applications. Independent agents can also have distinct profiles and presences within applications. For example, you could deploy different agents for different departments, each with its own name, identity, and access model.
Memory
Agents start by learning standard operating procedures from the company data and documentation that they have access to. They can extract key runbooks, and develop an initial picture of how to answer questions, unblock, or execute next steps. After that, they will learn from every interaction they have — like any other teammate — extracting best practices and ensuring that they don't make the same mistake twice.
Proactivity
Independent agents are designed to act without being asked. They'll suggest solutions, tag in key stakeholders, and serve as proactive companions for key processes. They also proactively share key learnings in regular summaries, so teams can make the most of critical insights and trends like on-call errors.
Accountability
None of this works if the agent can't be trusted, which is why accountability is built in rather than bolted on. You can open any run and see exactly what the agent did and why, with the ability to examine every tool call and every decision. They inherit Glean's security and governance model, and run inside your existing Glean tenant, with the same region and isolation as the rest of your enterprise data. Every agent has an emergency stop button that allows admins to disable it everywhere if something goes wrong.
Transforming engineering on-call with an AI coworker
The first of those agents is our OnCall Assistant in Slack. Here's a look at it in action:
We've already been using this assistant internally at Glean, and it has been a terrific AI coworker for our engineers. When an alert fires, the agent reads the escalation, pulls the relevant context, pursues multiple root-cause theories simultaneously, drafts a fix for review, and tags the right owner. By the time a human steps in, the first thirty minutes of tedious work are already done.
As the escalation progresses, OnCall Assistant serves as an AI coworker by pulling key information from docs and messages, analyzing logs, and suggesting the right solution as situations evolve and more context is gained. All the while, it learns from your company's procedures and its interactions to improve over time.
The future of agents at work is collaborative
Independent agents are AI coworkers that work hand-in-hand with your teams in true multiplayer fashion to execute critical processes together. They bring organizational knowledge directly to where your teams work so they have the right data, specialized expertise, and insights to rapidly move projects forward. We're excited to see what comes next as we bring additional specialized AI coworkers to market and explore how customers partner with their OnCall Assistant, starting in beta.
