Why we're building mantle. Read more
mantle

Every AI agent should know your business.

Mantle connects your data, resolves entities across every system, and serves quality-scored context — in one MCP call. No ETL, no glue code.

Get startedView documentation

Built by big tech data platform engineers.

How it works

Three steps from your data to an agent that actually knows you.

No ETL. No copies. No glue code. Connect once — every agent shares the same trusted context.

01

Connect

Plug in the sources your team already uses: Drive, Gmail, Slack, and Dropbox — with Postgres, S3, and more available on request. mantle reads in place. No ETL, no raw file copies, no sync jobs.

Google Drive
connected
Gmail
connected
SSlack
connected
Dropbox
connected

02

Understand

mantle resolves the same person, project, or customer across every system. Then links the records and builds one knowledge graph: the picture of your business no single tool has.

Raw — 4 sources, 4 names

Drive“John Smith”
Gmail“jsmith@acme.com”
Slack“JSmith”
Dropbox“J. Smith”

Resolved — 1 entity

entity_4f3a2c

John Smith

DriveGmailSlackDropbox
4 sources merged · 0 conflicts

03

Deliver

Your agent asks one question in plain English. mantle returns reasoned, quality-scored context over MCP, with source attribution for every claim. One call, every source.

Q“How is the Acme relationship going?”

Acme Corp is an active customer with 2 open support threads in Slack, 47 email exchanges in Gmail, and 12 documents shared via Drive. Relationship is strong — no unresolved escalations detected.

DriveGmailSlack
8 resultsscore 0.83

The bottom line

With Mantle vs without.

01

One call vs ten integrations.

Replace a sprawl of bespoke connectors and ETL jobs with a single MCP endpoint. Every agent in the org speaks to the same source of context.

02

Entities resolved, not duplicated.

Without Mantle, the same customer is four records in four systems. With Mantle, they’re one entity with four aliases — your agent stops splitting the truth.

03

Quality scored, not best guess.

Every fact carries a confidence and a citation. Your agent knows what to trust, what to verify, and never has to hallucinate the gaps.

Zero-copy architecture

Your data never leaves its source. We read in place.

Encrypted in transit & at rest

TLS 1.3 + AES-256 via Azure infrastructure.

We never train on your data

Your data is never used to train AI models. Ever.

You own your data. Always.

Delete anytime. No questions asked.

Full details in our security overview.

View security page

Built for two scales

For your team. For your work. Same mantle.

FOR ENTERPRISE

One context layer for every agent in your org.

Connect Salesforce, Snowflake, Slack, and the rest of your stack. Every internal agent (sales copilot, support assistant, ops automation, data lookup) shares the same trusted context. Permissions, audit logs, and entity resolution come standard.

Shared knowledge graph across teams
Source-level access controls preserved
Quality-scored context with full provenance
MCP-native, works with any agent framework
Talk to us about enterprise

FOR YOU

Your Claude, Cursor, or Code, with the context of your whole work life.

Connect your Gmail, Drive, Slack, calendar, and notes. Ask Claude what you committed to last week. Ask Cursor about that doc from three months ago. Your agent reads your data — nobody else's.

Personal data sources: Gmail, Drive, Slack, calendar
Read-only by default, fully revocable
Works with Claude Desktop, Cursor, Claude Code
Your tokens, your account, your control
Sign in to get started

MCP-Native

Plug in any MCP-compatible client.

mantle speaks the Model Context Protocol, the open standard from Anthropic. Connect any client. No adapters, no glue code, no vendor lock-in.

Claude
GChatGPT
Gemini
Cursor
VVS Code
Claude Code
Windsurf
LangChain
CrewAI
AAutoGen
+Custom agents

Data Sources

Reads your data where it lives.

Zero-copy connectors that read in place. No ETL, no duplication, no copies.

Available today

Google Drive
Gmail
SSlack
Dropbox
Notion

Available on request

PostgreSQL
Google Cloud Storage
S3Amazon S3

Coming soon

SSalesforce
Jira
HubSpot
GitHub
BigQuery
Confluence
Snowflake
4connectors live todayDrive · Gmail · Slack · Dropbox
<200msp95 latencyAt any scale
Zero‑copyarchitectureRaw files never leave your source

Comparison

How mantle stacks up.

Other context layers retrieve documents. mantle resolves who and what they're about, then ranks them by quality before serving.

FeaturemantleSupermemoryMem0Zep
Cross-source entity resolutionmantle onlyPartial
Passage-level chunking + LLM summariesmantle only
Confidence-scored factsmantle only
Knowledge graphPartial
Multi-hop graph traversal
Enterprise data connectorsPartial
No raw file duplication
MCP-nativePartialPartial
Sub-200ms p95 latencyPartial
Live database queriesPartial
Self-hostablePartial

Comparison reflects publicly documented features as of 2026. “Partial” denotes in-progress or limited availability. Get in touch if anything's out of date.

// The contrast

Without Mantle vs with Mantle.

Three concrete differences between today's agent stack and what Mantle delivers. No vague claims — just the gap in plain English.

01

Retrieval

Without Mantle

Stitched-together RAG.

Agents stitch together stale chunks from Drive, Notion, Slack, and your CRM — and hand back a confident answer that's half-hallucinated, half out of date.

With Mantle

Federated knowledge graph.

Agents query a typed, federated knowledge graph where Acme in Salesforce, Acme in the inbox, and Acme in the deck are one entity — resolved, time-aware, queryable.

02

Setup

Without Mantle

A pipeline per source.

Every new source means a new RAG pipeline, new permission glue, new vector store. The graph that lives in your head never reaches the agent.

With Mantle

Connect once.

Sources stay where they live. No copying, no re-indexing, no drift — the graph updates the moment your business does.

03

Trust

Without Mantle

No provenance, no ACLs.

No provenance. No timestamps. No ACLs. When the answer matters — to a customer, a regulator, a board deck — nobody can tell where it came from or who was allowed to see it.

With Mantle

Cited and permissioned.

Every fact carries a source, a timestamp, and an ACL gate. Your agents cite their work, respect who's asking, and never leak what a user wasn't supposed to know.

// Use cases

One MCP endpoint. Build anything.

Mantle is an MCP server, not a SaaS app to learn. Any MCP-compatible agent — Claude, Cursor, ChatGPT, your own — gets entity-resolved, quality-scored context across every connected source.

search_entities("Acme Corp")

Customer support copilots

Resolve the customer across CRM, ticketing, and Slack. Surface every open ticket, deal status, and recent escalation before the agent responds — across systems that don't share IDs.

get_entity_relationships("ent_acme_corp")

Sales pre-meeting briefs

One call returns the full relationship graph: emails, calls, contracts, product usage — with verbatim evidence and source attribution. Brief any sales agent in seconds.

semantic_search("auth service ownership")

Engineering onboarding

Help new hires find the canonical design doc, current owner, and recent decisions across Notion, Drive, GitHub, and Slack — without learning where each team hides things.

semantic_search("data retention clause")

Compliance & legal review

Search clauses across thousands of contracts, policies, and DPA addenda by meaning, not exact wording. Provenance-tracked so legal can verify the source.

search_entities("Project Aurora")

Personal context across your stack

Ask Claude or Cursor what you committed to last week. Pull from Gmail, Drive, Slack, and calendar — across your personal and work accounts, all read-only.

traverse_graph("ent_acme_corp", depth=2)

Analyst & research agents

Walk the knowledge graph two hops out from any entity. Surface every related contract, stakeholder, and decision — structured and unstructured context composed in a single call.

Why now

MCP is consolidating
as the integration standard.

A year ago, every team building an AI agent wrote its own adapter for every data source. Then the Model Context Protocol arrived, and the major agents and frameworks adopted it. Integration is no longer a bespoke problem.

The bespoke problem is one level up. Once every agent speaks the same protocol, the value moves to whatever sits behind it. A generic MCP server wrapping a vector DB isn't enough. Agents need resolved entities, scored context, and relationships across every source, delivered in a shape a model can actually act on.

That's the layer we're building. MCP made it a standard interface. mantle makes it a standard substrate.

What this means

Agents that know
stop guessing.

The wrong Acme

Your sales agent briefs the prospect instead of the existing customer. Same company name, two records, no resolution. mantle collapses them into one entity before the agent ever responds.

The missing escalation

Your support agent answers confidently — without knowing about the Slack thread from last Thursday. mantle indexes that thread and surfaces it with the ticket, every time.

The unnamed owner

Your engineering agent asks who owns the auth service. The answer is split across Jira, Drive, and Slack. mantle walks the relationship graph and returns one name, with the source.

For developers

Five minutes. One call. Any stack.

Drop your MCP endpoint into Claude, Cursor, ChatGPT, or any agent framework — and you’re live. No SDK, no migration, no glue code, no ETL pipelines to babysit.

  • Works with every MCP-compatible client out of the box.
  • From signup to first answer in under five minutes.
View Documentation
// Type-safe client against the live OpenAPI spec — no SDK required.
// Generate types: npx openapi-typescript https://api.mantleai.dev/openapi.json -o mantle.d.ts
import createClient from "openapi-fetch"
import type { paths } from "./mantle.d.ts"
const client = createClient<paths>({
baseUrl: "https://api.mantleai.dev",
headers: { Authorization: `Bearer ${process.env.MANTLE_API_KEY}` },
})
const { data } = await client.POST("/v1/search", {
body: { query: "Full relationship with Acme Corp", limit: 10 },
})
console.log(data?.[0]?.score) // → e.g. 0.91

Works with everything

TypeScriptPythonREST APIClaudeClaude CodeCursorChatGPTVercel AI SDKLangChainLangGraphCrewAIOpenAI SDKMastraZapiern8nPipecat

Get started

Your AI deserves
deeper context.

Sign in to Mantle and connect your first source in under five minutes. Or talk to us about an enterprise rollout.

Sign inTalk to us about enterprise

By signing in, you agree to our Terms and Privacy Policy.

Contact

Let's talk.

Whether you're evaluating Mantle for your team, scoping an enterprise rollout, or have a product or security question, we'd love to hear from you.

General Inquiries

hello@mantleai.dev