Glean started in early 2019 as enterprise search and has grown into an enterprise AI platform connected to a company's internal data and systems. When its customers asked for agents that could complete work, not just research it, Glean needed actions across enterprise systems and chose to partner with Composio rather than build that layer itself.
Customers wanted agents that finish the job
Glean's first product was search for employees inside their own company, built on transformers before the industry had a name for semantic search. As the AI models improved, the product moved from surfacing links to answering questions to doing work on a user's behalf.
Once Glean shipped agentic capabilities such as deep research and analysis, Arvind says the natural question from every customer was whether the agent could complete the full cycle of a task and save the result back into their systems. That required actions.
Arvind describes why actions are harder than they look: a quick system that executes a REST API call is easy, but doing it correctly with the right parameters, across tasks that need several API calls, runs into lots of complications, and for a long time LLMs were not good at it.
A partner-first strategy, applied to actions
Glean builds where it is uniquely strong, including search, deep knowledge and context graphs, and an understanding of how a business works, and partners for everything around that. Arvind puts the Composio partnership in the same category as Glean's partnerships with model providers: great technology made available as a platform that Glean would rather leverage than reinvent.
Composio was deeply focused on the one specific area of actions, and Glean expected that focus to translate into speed and depth. The relationship also predates the partnership: Composio co-founder Karan was one of Rubrik's early engineers when Arvind ran that company, which Arvind says made partnering a clear choice.
From answers to agents that act inside enterprise systems
Glean's platform now reads information and takes actions across a customer's enterprise systems, and that same platform is available to customers to integrate into their other AI tools. Many customers connect Claude Code or Cursor to the Glean MCP gateway to bring company context into those tools.
On measuring value, Arvind points to departmental metrics: one of the largest telcos in the US runs its customer care function on Glean and, as of last year, had already cut time to resolve a case by 48%. His view is that AI is now ready for all knowledge work, and that success depends on following the right approach and getting the right tools in place, including an actions platform.
