The Quiet Challenger Nobody Is Talking About
Glean built its reputation as the enterprise search company – the tool that promises to find anything across your company’s scattered apps and data. Dust is building something that makes that promise look like a consolation prize.
What Dust Actually Does (And Why It Matters)
Dust positions itself as an AI assistant layer for enterprise teams, but that description undersells what the product actually does in practice. Where Glean is fundamentally a search interface – you ask, it retrieves – Dust is built around the idea that knowledge retrieval should be the beginning of a workflow, not the end of one. The company lets teams build custom AI agents that connect to the same data sources Glean indexes, but instead of returning a list of documents, those agents reason over the content and take action on it.
The architecture matters here. Dust was founded by former Stripe and Google engineers, and the product reflects an engineering-first philosophy that prioritizes configurability over polish. Teams can define what their agents know, how they respond, what tools they can access, and what guardrails they operate within. That kind of granular control is not something Glean was designed to offer, because Glean’s core assumption is that search – done well enough – is the primary job to be done. Dust’s core assumption is that retrieval is just the first step.
The company has been growing quietly among technical and operations teams at mid-market companies that have already tried Glean and found it useful but limited. The complaint is almost always the same: Glean is good at finding things, but it doesn’t do anything with what it finds. You still have to take the retrieved information and carry it somewhere else, into another tool or another conversation. Dust closes that gap by treating the agent layer as the product, with retrieval as infrastructure underneath it.
This is not a minor distinction. Enterprise software buyers have been burned enough times by single-feature tools that they are increasingly asking whether a product can grow with their needs. Glean’s answer to that question requires integrating with other platforms. Dust’s answer is to absorb the use cases that would otherwise require those integrations in the first place.
Where Glean Is Vulnerable
Glean’s strongest selling point has always been its connectors – the company has spent years building integrations with Salesforce, Notion, Confluence, Slack, Google Drive, and dozens of other enterprise platforms. That connector moat is real, and it took significant engineering time to build. The problem is that the moat protects the retrieval layer, not the value delivery layer. And the value delivery layer is where enterprise buyers are now spending their budget.
Search as a category is getting squeezed from two directions at once. From below, AI-native features are being baked directly into the tools enterprises already use – Notion AI, Salesforce Einstein, Microsoft Copilot – making the case for a standalone search tool harder to make at the C-suite level. From above, agentic platforms like Dust are arguing that retrieval without reasoning is an incomplete solution. Glean is caught in the middle, and its product roadmap suggests it knows this, with the company increasingly pushing its own AI assistant features as a response. But bolting an assistant layer onto a search product is architecturally different from building the assistant layer first.
Dust’s approach also plays well with the way enterprise AI buying is actually happening right now. Procurement teams are consolidating tools, not adding them. When a head of operations can demonstrate that Dust handles the knowledge retrieval, drafting, summarization, and routing that previously required three separate tools, the ROI conversation becomes significantly easier. Glean, by contrast, tends to be positioned as an addition to the stack rather than a replacement for part of it. That positioning is a liability when budgets are under pressure.
The customer profile Dust is targeting also tells a story. The company’s early adopters skew toward companies in the 200-to-2000 employee range – too large to operate informally but not large enough to have dedicated IT teams managing enterprise search infrastructure. These are exactly the companies Glean has been courting aggressively, and they are the ones most likely to ask whether an agent platform could eliminate the need for a separate search license entirely.
There is a pattern developing in enterprise software where a new entrant doesn’t attack the market leader directly, but instead reframes what the category is actually for. Linear did this to Jira by arguing that issue tracking should feel like a product you want to use, not one you’re required to. Dust is making a similar argument about enterprise knowledge: that search was never really what teams needed – action was.
The Question Glean Has to Answer
Glean is not standing still. The company has raised significant venture funding and has been shipping product aggressively, including its own Glean AI assistant that attempts to layer reasoning on top of search results. The challenge is that this assistant is constrained by the same assumption that built the company: that the search index is the primary asset, and the assistant is a feature on top of it. Dust inverts that relationship entirely, and inverting it is easier than working around it.
What gives Dust’s trajectory its real weight is that enterprise buyers are asking a different question than they were two years ago. In 2022, the question was “can we find our internal knowledge faster?” In 2024, the question is “can AI actually do something with our internal knowledge?” Those are not the same question, and they don’t have the same answer. Glean was built for the first question. Dust was built for the second – and in enterprise software, the company that solves for where the buyer is going tends to outlast the one that solved for where they were.
