
Your company's knowledge is scattered: chats, calls, documents and repos
Why what your company knows is split across chats, calls, documents and repos, what searching costs, and what it takes to bring it together with citations.

Glean is an enterprise AI platform with 275+ connectors and contract-only pricing; MemoryFirst is cited memory over MCP for your agents. Facts as of 2026-11-14.
A company of 3,000 people and a six-person studio share the same underlying problem: what they know is scattered across email, chats, documents and heads, and nobody knows where to look. Glean and MemoryFirst attack it from opposite ends. Glean is an enterprise AI platform built for large organizations. MemoryFirst is cited memory that the agents your team already uses connect to over MCP. We compare them using data from their public websites, accessed on 2026-11-14.

Glean describes itself as "the Enterprise AI platform connected to your enterprise's data" and advertises "275+" connections to work apps (Slack, Google Drive, Jira, Confluence, SharePoint, GitHub, Salesforce, among others). Named customers include Rivian, Samsung, Booking.com, Zillow, Databricks and T-Mobile. Its documentation describes two deployment models: Glean Hosted (SaaS, on GCP) and Customer Hosted, where Glean operates its isolated tenant inside your own GCP or AWS account.
MemoryFirst ingests email (Gmail, IMAP), calendar, Drive, Dropbox, OneDrive, Slack, files and events through an API; indexes them with dense retrieval, BM25 and a graph; and answers with citations (page, minute, message). It is consumed through a remote MCP server with 15 tools, a REST API or a stdio server. We do not have 275 connectors: we have the ones just listed, and others (Linear, Sentry, audio) are a plan, not a fact.
| Aspect | Glean | MemoryFirst |
|---|---|---|
| Focus | Enterprise AI platform: search, assistant, agents | Cited memory for agents and teams |
| Connectors | "275+" according to its website | Gmail, IMAP, Calendar, Drive, Dropbox, OneDrive, Slack, files, events API |
| Permissions | Queries enforce the user's permissions through its knowledge graph | Owner, member and viewer roles; we do not mirror each source app's permissions |
| MCP | Managed remote MCP server (search, chat, document reading and more); also acts as a host for MCP servers | Remote MCP server (OAuth 2.1 or API key) and stdio |
| User memory | Saved and learned memories, per user, facts and preferences only, not document content | Organization memory, temporal graph with contradictions resolved |
| Citations in answers | We could not verify this on the pages we read | Yes, span_ref per source type |
| Deployment | Hosted by Glean or in your own cloud (GCP or AWS) | Managed service only; no self-hosting |
| Certifications | Its site lists ISO 42001, HIPAA, TX-RAMP Level 2, SOC 2 Type II, ISO 27001 and GDPR, without dates or scope | We claim none |
| Public pricing | No: on request only | Yes: Free, Pro and Team |
Glean. Its pricing page (accessed 2026-11-14) shows no plans, amounts or minimums. Its licensing documentation describes the "Enterprise Flex" model: seats licensed per user per month, plus a pooled, pay-per-use allowance of "FlexCredits" for advanced features (premium models, agent runs, meeting notes); client APIs consume FlexCredits. There are annual site-wide add-ons (Protect+ and Premium Support). The documentation contains no amount in dollars or euros. Per-user figures circulate on third-party blogs, often from competitors; they do not come from Glean and we do not repeat them.
MemoryFirst. According to our pricing page: Free (1 project, memory over MCP and Claude Code hooks, 30-day retention, 1 seat); Pro, 12 EUR a month billed annually (144 EUR a year; monthly costs 20% more), with unlimited projects and retention, all connectors, a temporal graph with citations and the morning brief; Team, 39 EUR a month billed annually (468 EUR a year), with 5 seats and shared memory. VAT included. Each extra Team seat costs 9 EUR a month billed annually (10.80 EUR month to month).
Pros: very broad application coverage; user permissions applied on every query; option to run it in your own cloud; certifications declared on its site; managed MCP and agents; large reference customers.
Cons: no public price and no purchase without talking to sales; the seats-plus-usage-credits model forces you to estimate consumption; its MCP docs mention usage-based pricing and beta or general-availability stages we could not clarify; built for large organizations, and we did not see it aimed at five-person teams (not verified).
Pros: public, low price; sign-up with no rollout project; verifiable citations; open MCP that any agent can consume; database, embeddings and files in the EU; export and verifiable erasure.
Cons: far fewer connectors; we do not mirror each source app's permissions; no certifications to announce; default inference goes to Cerebras (US) unless you bring your own key; no self-hosting; tenants separated by query filters plus an optional second barrier, not yet by separate databases.


The AI memory market in 2026 places both. If your doubt is between agent memory and company memory, read the comparison with Letta and how to choose a memory layer. For what this means in practice, see what a company brain is, the company brain page and the plans.
All accessed on 2026-11-14. Anything not listed here could not be verified.
Start free with one project and move to Team when several people need the same memory.
Start free
Why what your company knows is split across chats, calls, documents and repos, what searching costs, and what it takes to bring it together with citations.

A simulated scenario: six people on Claude Code and Cursor, architecture decisions that get lost, and a shared memory over MCP, with illustrative numbers.

A simulated scenario: an agency with 12 clients, promises scattered across emails and calls, and what client separation MemoryFirst offers and what it does not.