Glean — enterprise search and work AI assistant across company knowledge

Investment Memo: Glean

Company: Glean (enterprise search & AI work assistant) Stage evaluated: Hypothetical Series A Author: [Partner], [Fund]

---

Thesis

Every knowledge worker loses hours weekly searching across Slack, Drive, Jira, Salesforce, Confluence, and email. The data is fragmented; the org chart of information is invisible. Glean's bet is that the company that indexes all enterprise knowledge with permissions-aware retrieval becomes the natural interface layer for work — first as search, then as the AI assistant every LLM application needs to be grounded in. If that's right, Glean isn't a search tool; it's the retrieval and permissions infrastructure for enterprise AI, a position with system-of-record-like defensibility and seat-based pricing across the entire white-collar workforce. That's a fund-returner shape: horizontal, per-seat, sticky, and sitting on the choke point (the index) that every AI agent will need.

The founder quality matters here: Arvind Jain co-founded Rubrik and was a distinguished engineer at Google Search. This is arguably the single best-matched founder-problem pairing in enterprise software — a rare "why this team" that's actually dispositive.

Product & wedge

The wedge is unified enterprise search: 100+ connectors, a knowledge graph of people/documents/activity, and — critically — native permission enforcement at query time (you only see what you're entitled to see). That permissions layer is brutally hard, unsexy, and the real moat; it's why naive "RAG over your Slack" startups die in security review.

From search, Glean expanded into an AI assistant (Glean Chat/Assistant) that answers questions grounded in company knowledge, and now an agent platform letting enterprises build workflows on Glean's index. The sequencing is right: search earns the data access and trust; the assistant monetizes it; agents deepen lock-in.

Market & competition

TAM framing: ~1B knowledge workers; even $30–40/seat/month across mid-market and enterprise implies a multi-tens-of-billions market. The realistic near-term market is the enterprise search + AI assistant budget line, likely $10B+ and growing as AI budgets form.

Competition is severe and comes from three directions:

1. Platform incumbents: Microsoft Copilot (bundled with M365, already indexes via Graph), Google (Agentspace/Gemini for Workspace), and OpenAI/ChatGPT Enterprise with connectors. Microsoft is the existential one — distribution plus data gravity.
2. Direct startups: Coveo (public), Elastic-based builds, Hebbia, Dust, Onyx (open-source), Perplexity Enterprise, Sana, Moveworks (adjacent, acquired by ServiceNow).
3. App-level assistants: Notion AI, Slack AI, Atlassian Rovo — each trying to be "search across your stuff" from within their own suite.

Glean's counter-positioning: neutrality (indexes everything, favors no suite), best-in-class connectors and permissions, and model-agnostic LLM orchestration. Neutrality is a real advantage in heterogeneous enterprises — but only if the product gap over Copilot stays wide.

Traction & business signal (publicly known)

• Raised through Series F; reported ~$260M in 2024 at ~$4.6B valuation and a subsequent 2025 round reported at ~$7.2B. (Amounts per press reports; exact terms unknown.)
• Publicly reported to have crossed $100M ARR (company statements, early 2025), with claimed rapid growth; exact growth rate, NRR, and burn: unknown.
• Marquee logos cited publicly: Databricks, Canva, Duolingo, Reddit, Booking.com, among others.
• Investors include Sequoia, Kleiner Perkins, Lightspeed, General Catalyst, ICONIQ — strong signal, though also priced in.
• Gross margin, sales efficiency, seat expansion rates, churn: unknown.

At the hypothetical Series A frame (rewinding to ~2021): the signal then was Jain + founding team ex-Google Search/Facebook, early design-partner deployments, and unusually deep connector/permissions engineering pre-launch. Revenue would have been minimal — this was a team-and-thesis bet.

Risks (the three that kill the deal)

1. Microsoft bundles you to death. Copilot ships "free enough" inside E5, indexes M365 natively, and steadily adds third-party connectors via Graph. If Copilot becomes 80% as good, CIOs won't pay $40/seat incremental for Glean. The honest counterargument — Microsoft's cross-suite indexing is mediocre and enterprises are heterogeneous — is true today but is a wasting asset.
2. The index gets commoditized by MCP/agent protocols. If standardized connector protocols (e.g., MCP) let any LLM query SaaS apps directly with federated permissions, Glean's painstakingly built index becomes a feature, not a moat. The retrieval layer could be disintermediated by the model providers themselves — OpenAI and Anthropic both clearly want this position.
3. Value attribution and seat economics. Enterprise search has historically been a graveyard (Autonomy, FAST, Google Search Appliance) because value is diffuse and hard to attribute; renewal conversations are vulnerable in budget cuts. If Glean can't convert from "nice search" to "agents doing measurable work," NRR compresses and the per-seat model caps out below the thesis.

Verdict

Invest at Series A. This is a top-decile team attacking the correct choke point (permissions-aware universal index) with correct sequencing, in a category where the winner plausibly becomes the enterprise AI interface layer. The competitive risk from Microsoft is real but symmetric to every horizontal enterprise bet, and Glean's neutrality plus engineering depth is the best available hedge.

The one question I need answered first: In accounts running both Copilot and Glean, what is Glean's daily active usage and query share after 6 months — and does it hold when Copilot is free? If Glean sustains dominant usage head-to-head against a bundled incumbent, the moat is real. If not, this is a great product waiting to be flattened.

Thesis

Glean could be a fund-returner because it sits at the intersection of two unusually large budgets: enterprise knowledge management and generative AI. Most companies have fragmented information across Google Drive, Slack, Salesforce, Jira, ServiceNow, SharePoint, email, and internal applications. The resulting cost is substantial: employees waste time finding information, duplicate work, and route routine questions to scarce experts. Glean’s ambition is to become the permission-aware interface to all company knowledge—and increasingly, the system that takes action across it.

The venture case is not simply “enterprise search plus an LLM.” If Glean can establish a trusted company-wide knowledge graph, maintain document-level permissions, and become embedded in daily workflows, it may achieve high retention, broad deployment, and expansion from search into assistants and agents. The strategic value could be significant: Microsoft, Google, Salesforce, ServiceNow, and Atlassian all have reasons to own this layer. A category leader could plausibly become a large independent software company or an acquisition target.

The counterpoint is that this is a valuable feature set surrounded by platform giants. The investment only works if Glean’s cross-system relevance, deployment speed, and neutrality are meaningfully better than bundled alternatives.

Product & wedge

Glean began as enterprise search: connect to a company’s applications, index content, respect underlying access controls, and return relevant answers. Its product has expanded into an AI assistant that synthesizes answers, cites sources, drafts content, and supports customized workflows or agents.

The wedge is practical rather than flashy. Enterprise search has historically failed because results were stale, disconnected from context, or exposed sensitive information. Glean’s differentiation is intended to be its enterprise knowledge graph, which links people, teams, documents, projects, and business concepts, combined with broad connectors and permissions-aware retrieval. A new customer can start with search and question-answering, then expand into department-specific assistants and automated actions.

This wedge also creates a potential data and workflow moat. Usage can improve relevance, while integrations and organizational context make replacement increasingly painful. However, the moat is only durable if Glean consistently retrieves the right information and earns enough trust for employees to use it for consequential work.

Market & competition

The market is large but difficult to define. Enterprise search, knowledge management, collaboration software, and AI productivity tools collectively represent tens of billions of dollars in software spend. The immediate customer budget may come from IT, employee productivity, search, or business applications rather than a single established category.

Competition is intense:

• Microsoft Copilot has privileged access to Microsoft 365, Teams, SharePoint, and Graph, plus enormous distribution.
• Google Gemini for Workspace has a similar advantage inside Google-centric companies.
• Atlassian Rovo combines enterprise search and AI over Jira, Confluence, and connected applications.
• ServiceNow, Salesforce, and Workday are embedding assistants into systems of record.
• Notion, Guru, Box, Coveo, and Elastic compete in knowledge, search, or AI retrieval.
• Internal builds using foundation-model APIs, vector databases, and existing data platforms are credible for sophisticated enterprises.

Glean’s opportunity is to be vendor-neutral and cross-application, particularly for companies with heterogeneous stacks. Its challenge is that neutrality may be less valuable than the distribution and data access enjoyed by platform vendors.

Traction & business signal

Publicly known signals are strong, though not all are independently verified. Glean was founded in 2019 by former Google executives, including CEO Arvind Jain. It has raised more than $350 million in disclosed funding. In March 2024, Glean announced a $200 million financing at a reported $4.6 billion valuation.

The company has publicly described rapid growth, including more than 1,000 customers and annual recurring revenue exceeding $100 million; the exact current ARR, net retention, gross margin, customer concentration, and free-cash-flow profile are unknown. Public customer references include large enterprises such as Databricks, Grammarly, and other recognizable technology companies, but the full customer cohort, average contract value, implementation period, and usage-based expansion economics are unknown.

The financing and customer count indicate genuine commercial demand rather than a laboratory product. They do not, by themselves, establish durable retention or attractive returns at the reported valuation.

Risks

1. Platform displacement. Microsoft and Google can bundle much of Glean’s functionality into products customers already own. If Copilot and Gemini become “good enough,” procurement may reject a separate search layer. This is the most direct existential risk, because distribution and data access can matter more than model quality.

2. Trust and retrieval failure. Enterprise AI is only valuable when answers are accurate, current, and properly permissioned. A serious data leak, confidently wrong answer, or inability to explain sources could halt adoption. Conversely, excessive security constraints can make the product feel incomplete. Glean must solve both technical correctness and organizational trust at scale.

3. Weak or expensive expansion economics. Search may be broadly useful but difficult to monetize per seat, while agents may require substantial services, support, and model-inference costs. If customers buy pilots but do not deploy company-wide, Glean could have impressive logos and bookings without durable net retention or software margins.

Verdict

Invest at a hypothetical Series A, subject to disciplined valuation and evidence of repeatable enterprise deployment. The problem is painful, the product can become a control point for work, and the cross-application position gives Glean a credible opening against suite vendors.

The one question I would need answered first is: What are gross dollar retention, net revenue retention, and time-to-company-wide deployment for customers acquired at least 18 months ago—and how do those metrics compare with customers using Microsoft or Google’s native AI products? If Glean can demonstrate durable, expanding usage despite platform competition, it merits conviction; if not, it is likely an expensive feature rather than a fund-returning company.