One of the things we find ourselves repeating in almost every Glean conversation is this: agents are built by business teams, not developers.
It sounds like marketing. We understand the scepticism. But having implemented Glean across multiple Australian organisations, we can confirm it's genuinely how it works — and it changes the economics of AI automation entirely.
What is a Glean agent?
A Glean agent is an automated workflow that can do real work on your behalf. Not just answer questions — actually take action.
A few examples we've seen teams build:
- An HR onboarding agent that answers new starter questions by pulling from policies, handbooks, and internal wikis — without anyone in HR needing to respond manually.
- A sales enablement agent that compiles relevant case studies, pricing, and competitive positioning based on the specific prospect a rep is about to meet.
- An IT helpdesk agent that resolves common support tickets by walking employees through solutions sourced from knowledge base articles and past tickets.
- A compliance agent that checks draft documents against regulatory requirements and flags potential issues before review.
None of these were built by developers. They were built by the people in HR, sales, IT, and compliance who understood the work.
Why business-built matters
In most organisations, the bottleneck for AI adoption isn't ideas. It's implementation capacity.
Every team has a list of things they wish AI could do. But if every one of those ideas requires a developer to build, test, and deploy, the list grows faster than the team can execute. Ideas wait in backlogs. Enthusiasm fades. The organisation moves slowly while the market moves fast.
Glean's approach removes that bottleneck. The people with the domain knowledge — the people who know what questions customers ask, what processes slow things down, what information is hard to find — are the same people who build the agents.
This is what we mean when we say: spend time using AI tools in your business, not developing them.
How it works
Building a Glean agent follows a straightforward process:
1. Define the purpose
What should this agent do? What questions should it answer? What actions should it take? Start narrow. The best agents solve one specific problem well.
2. Point it at the right knowledge
Glean agents draw from the knowledge already in your connected systems. You tell the agent which sources to use — specific Confluence spaces, SharePoint sites, Slack channels, or document collections. The agent uses Glean's knowledge graph to find relevant information.
3. Set the behaviour
You define how the agent should respond. What tone? What format? Should it cite sources? Should it escalate to a human in certain situations? These are instructions written in plain language, not code.
4. Test and refine
Ask the agent the questions your team actually gets. See how it responds. Adjust the instructions, add or remove knowledge sources, refine the scope. This iteration cycle takes hours, not weeks.
5. Deploy to your team
Share the agent with the people who need it. Monitor usage, collect feedback, and improve over time.
What makes this different from ChatGPT?
You could paste documents into ChatGPT and ask it questions. Some people do. But there are important differences:
- Security: Glean agents respect your existing permissions. They won't surface information the user shouldn't have access to. ChatGPT has no concept of your permission model.
- Currency: Glean agents draw from live, connected systems. The information is current, not a copy-paste from last week.
- Context: Glean's knowledge graph means agents understand relationships — who wrote what, which project it relates to, what's recent and relevant.
- Governance: You control what the agent can access, how it responds, and who can use it. There's an audit trail.
Where to start
The best first agent is one that solves a problem your team complains about regularly. The one that generates the most internal Slack questions. The process that requires someone to manually look things up and compile answers.
Start there. Build it in an afternoon. See what happens.
If you'd like help identifying the right first agent for your organisation — or you'd like to see what other Australian businesses have built — talk to the JOURN3Y team. We implement Glean and help teams build their first agents as part of every engagement.