Glean launches Tau desktop workspace, claims token-cost edge over Claude Cowork

Tau runs on an open-source harness and pairs Glean's enterprise context with whatever sits on the machine. Glean said the tool reduces what it calls botsitting, the time users spend hunting for the right files, feeding in background and steering a model through each step. Tau can plan multistep work, carry it out on a user's behalf and check its own output. File organization, document analysis and spreadsheet building are among the jobs it handles, with coding and engineering workflows as an early use case.

Cost efficiency runs through the rest of the announcements. Glean argued that keeping permission-aware company context available to a model avoids spending tokens on repeated retrieval calls that rebuild that context at runtime. An expanded Glean Intelligence adds centralized usage visibility along with administrative controls over AI spending. Administrators also get enhanced auto routing, letting them push work toward efficient, balanced or frontier models depending on the price-performance tradeoff they want.

Six other products and expansions came out of the conference. Glean Transform reads real activity signals to work out how work actually gets done, recommends which pieces AI should take on and measures the business impact. A proactive task management feature drafts follow-ups and prepares users for meetings before anyone asks, with email triage pulling company context into the queue and independent agents taking on longer-running work. Interactive dashboards are being expanded to blend structured data with tickets, conversations and documents, refreshing as the underlying information changes.

The remaining three cover collaboration, governance and security. Team chat brings several people into a shared thread or artifact with Glean. The company's AI Gateway is being widened to support more entry points, enforce restricted-topic policies through Glean Protect and extend governed Model Context Protocol access to organizational skills and personal memory. Context-aware threat detection draws on the Enterprise Graph to separate legitimate agent activity from data harvesting, exfiltration or destructive action, including cases where each individual step looks permitted.

Emrecan Dogan, chief product officer at Glean, said models keep improving and are increasingly interchangeable, while understanding of a given enterprise is not. Glean was designed around that in 2019, he said, and company context grows more valuable as AI moves from answering questions to doing the work, “because it tells AI what matters, what should happen next and how to act within the realities of the business.” Two customers appear in the announcement. Rhonda Baldwin, chief information officer at LaunchDarkly, credited the platform with letting her teams scale AI faster without giving up governance. David Lee, director of people and places enablement at LegalZoom.com Inc., said recreating the same depth of context on other platforms would take a huge number of tokens.

Availability splits three ways. Glean Intelligence, the AI usage controls and memory via MCP are generally available. Beta covers independent agents, the AI Gateway, auto routing, team chat and the dashboard refresh. Tau is not out yet, and neither are Glean Transform, threat detection, task management, email triage, a meeting coach or skills via MCP. The company also launched a global partner network, spanning resellers, systems integrators, cloud providers and private equity firms across four regions.

Glean is venture capital-backed and last raised $150 million in a Series F round at a valuation of $7.2 billion in June 2025, according to SiliconANGLE.