The Quiet Challenger Nobody Saw Coming
Glean built its reputation as the enterprise search layer that actually works – a tool that indexes your Slack, Notion, Salesforce, and Gmail and lets employees find things without digging through six tabs. It raised hundreds of millions of dollars on that promise and landed contracts with some of the largest companies in the world. But a Paris-founded startup called Dust is now eating into that workflow base, not by competing on search, but by going one layer deeper into how enterprise teams actually use AI at work.
Dust does not pitch itself as a search product. It pitches itself as a platform for building custom AI assistants that are connected to a company’s internal data, tuned to specific teams, and governed by actual security policies. That distinction matters more than it sounds. Where Glean gives employees answers, Dust gives teams agents – and right now, the enterprise market is paying a premium to move from retrieval to action.
What Dust Actually Builds
The core product is a workspace where non-technical teams can create AI assistants without writing code. A customer success team might build an assistant trained on their support tickets, product documentation, and CRM history. A legal team might build one that surfaces relevant contract clauses before a negotiation. The assistants are not generic – they are built on the company’s own data, with permissions that reflect who should see what. That last part is where most enterprise AI tools break down, and Dust has made it a selling point rather than an afterthought.
Dust connects to the data sources Glean also indexes – Google Drive, Notion, Confluence, Slack, and others – but the workflow diverges sharply after that. Glean surfaces information. Dust builds agents that do something with it. A growing number of enterprise buyers are realizing those are not interchangeable products, even if they both start with a search bar. That distinction is what makes Dust a threat Glean cannot fully neutralize by simply adding an AI layer on top of its existing search infrastructure.
The Workflow Problem Glean Has Not Solved
Glean’s strength is breadth. It connects to almost everything and returns results fast. That is genuinely useful for a company trying to stop employees from re-asking questions that already have answers somewhere in the company’s knowledge base. But the product stops working well when the task requires synthesis, reasoning across multiple sources, or taking an action based on what was found. Those are the exact moments where knowledge work gets expensive and slow.
Dust is positioned to handle those moments. An assistant built in Dust does not just retrieve a policy document – it can read the document, compare it against a draft contract, and surface the specific clauses that need attention. That is a different category of output, and it is the kind of output that makes a department head willing to allocate real budget. The shift from search to agent-based workflows is not a minor product upgrade – it changes the ROI calculation for buyers entirely.
There is also a structural problem Glean faces as AI becomes more central to enterprise software. Glean is essentially a middleware product – it sits between employees and their tools and makes retrieval faster. But as AI agents become capable of working across those same tools directly, the middleware layer gets thinner. Why search for an answer when an agent can go get it, synthesize it, and act on it without a human in the loop? Dust is building for that world. Glean is still optimizing for the previous one.
The competitive pressure Dust is applying is not frontal. It is not running ads against Glean or positioning itself explicitly as an alternative. Instead, it is entering enterprises through individual team champions – a head of operations here, a VP of customer success there – and expanding from within. That bottom-up motion is the same playbook that helped tools like Linear displace entrenched project management incumbents by winning developers before procurement ever got involved.
The Security and Governance Angle
One of the most consistent complaints from enterprise IT teams about early AI tools was that they made data governance impossible. Tools trained on company data without clear permission controls created legal and compliance headaches that often ended pilot programs before they scaled. Dust built its permission architecture early, and it shows. The platform inherits the access controls from connected data sources, meaning an assistant built for the sales team cannot accidentally surface HR data that was never meant to cross that line.
That is not a flashy feature, but it is the feature that makes procurement meetings shorter. Enterprise security teams have become the unofficial gatekeepers for AI adoption, and any tool that cannot give a clear answer about data handling gets killed before it reaches a real rollout. Dust’s architecture speaks directly to that concern, which gives it an advantage in late-stage enterprise deals that a more capable but less governed tool would lose.
Where the Competition Gets Complicated
Glean is not standing still. The company has added AI-generated summaries, meeting recaps, and answer synthesis features that push it closer to what Dust does. But there is a difference between adding generative features to a search product and building a platform where teams can customize and deploy their own agents. Glean is doing the former. Dust is doing the latter. For buyers who just want smarter search, Glean remains strong. For buyers who want to build actual AI workflows, Dust is becoming the default conversation.
The enterprise AI market is also not winner-take-all. Large organizations routinely run five or six overlapping productivity tools across different teams. Dust can coexist with Glean in the same company – and in several cases, it already does. But coexistence has a ceiling. As AI budgets consolidate and IT leaders push for fewer vendors, tools that do more win contracts over tools that do one thing well. Dust’s platform model is better built for that consolidation than a single-purpose search product.
What makes the dynamic genuinely interesting is that Dust’s Paris origins and relatively quiet fundraising profile mean it has flown under the radar in a market where American AI startups with nine-figure raises dominate the conversation. It is not the loudest name in enterprise AI. But the teams adopting it are not switching back – and the use cases spreading inside those organizations are exactly the ones Glean was supposed to own.
