Does Sentra Make Claude Know Your Company? Sentra for Claude (2026)

TL;DR

Yes, Sentra makes Claude know your company: you connect Claude to Sentra's company memory over MCP, with one command in Claude Code (claude mcp add --transport http sentra https://api.sentra.app/mcp/) or a custom connector with the same URL in the Claude web and desktop apps, and Claude then answers from your company's facts with the permissions of the person using it. Sentra here is sentra.app, the organizational memory and company brain for teams and AI agents, not Sentra Inc (sentra.io), the data security company.

• Claude on its own knows what you tell it: its memory is per user, Projects hold files you upload, and Claude Code reads CLAUDE.md files in each repository.
• Sentra reads email, chat, meetings, documents, ERP, CRM, HRIS and BI, turns them into facts about people, accounts and projects with their source and time, and serves them to Claude through the Sentra MCP server.
• Claude sees only what the person it acts for may see, under each source's own permissions. A failed permission check counts as no access.
• On Terminal-Bench 2.1, a coding agent with Sentra scored 88.31% against an 83.37% baseline, at 72.6% lower model cost with 41.2% fewer tokens.

What "Claude knows your company" means

A Claude that knows your company can answer "who owns the Acme renewal and what did we promise them?" from the meeting, the email thread and the CRM record, cite each one, and give the same answer to everyone allowed to see it. Claude's own features do not do that by themselves. Claude memory for teams explains what persists and what resets: memory remembers one user, Projects remember one project, and nothing in either learns what the rest of the company decided.

Sentra supplies the company side. It does the work of resolving names, keeping sources and applying permissions once, when the data arrives, so Claude asks for the facts about an account or a person and gets a short, cited, current answer instead of a pile of raw documents.

How to connect Sentra to Claude

Claude Code

One command adds the Sentra MCP server:

claude mcp add --transport http sentra https://api.sentra.app/mcp/

Then run /mcp once in a Claude Code session to sign in and authorize. The trailing slash on the URL is required.

Claude web and desktop apps

Anthropic supports custom connectors over remote MCP on Free, Pro, Max, Team and Enterprise plans, with Free limited to one custom connector (support.claude.com). On an individual plan, add a custom connector named Sentra with the URL https://api.sentra.app/mcp/ and approve the Sentra sign-in. On Team and Enterprise plans an owner adds the connector for the organization and each member connects with their own account, which is what keeps every person's access their own.

The org-admin steps, including the setting that stops Claude asking every user to approve every tool, are in how to give Claude organizational memory over MCP.

Claude in Slack

Claude Tag, Anthropic's Claude in Slack for Team and Enterprise customers, works with the tools and data an admin chooses to connect, according to Anthropic's launch post. Our Sentra vs Claude Tag comparison covers how a Slack agent and a company memory fit together.

What Claude gets from Sentra

The Sentra MCP server groups its tools by what they read, and a caller sees only the tools it is entitled to (memory for agents):

• Search and fetch across meetings and entities
• Entities, the facts about an entity and its relationships
• Meetings, with summary and transcript
• A research agent that takes a natural-language question and returns a written answer with citations
• Playbooks, connected sources, stalled threads and follow-up load
• Email threads and raw email bodies, only with a separate email access permission

Search returns at most 20 results per call and each key is limited to 120 requests a minute, so one call cannot flood Claude's context.

Claude with Sentra vs Claude on its own

Permissions follow every fact

A fact in Sentra inherits the permissions of the system it came from, checked when someone asks, not when the data was loaded. For Slack, Linear, GitHub, Google Chat and Jira, access is checked against the source system itself, and any error counts as no access. Claude acting for a person gets exactly that person's access, and raw email bodies need a separate permission that no built-in role includes. Withheld facts are simply absent, with no hint that anything was hidden. See how Sentra enforces permissions on every fact.

Keeping what Claude reads current

Each fact keeps its source and the time the event happened, so Claude gets the newest facts first. When a source changes, Sentra rebuilds the facts from that source, so an edited document does not leave old facts behind. When sources disagree, Sentra answers from the newer one and says that the older one disagrees. Corrections people make in chat are kept as dated facts, and a later sync never overwrites them. See how Sentra keeps company memory current.

Why this makes Claude cheaper as well as better

Without a memory, Claude rereads raw documents and messages on every call and works out who is who and what is current each time. With Sentra it receives distilled facts, so it spends fewer tokens finding context and more on the task. On Terminal-Bench 2.1, across 445 trials, a coding agent with Sentra reached 88.31% mean reward against an 83.37% published baseline, at 72.6% lower model cost and with 41.2% fewer tokens (research).

To see Claude answer from your own company's memory, book a demo.

FAQ