One MCP endpoint. Build anything.

01 · Financial services

Pre-trade client intelligence in seconds.

Wealth managers, IB analysts, and trading desks need a single view across CRM, deal pipeline, research notes, market data, and client communication. Mantle resolves the client across every system, ranks the most relevant facts, and serves them to your agent — so the next call starts with full context, not a guess.

// Example query

mantle.get_entity_context("ent_westbrook_cap_4f3a", verbosity: "detailed")

// What comes back

{
  entity: { name: "Westbrook Capital", aum: "$2.4B", primary_contact: "Maren Vosh" },
  positions: { open: 14, avg_holding_days: 87, top_concentration: "muni_bonds" },
  recent_communication: [
    { type: "call",  duration_min: 32, summary: "Pricing pushback on muni tier" },
    { type: "email", subject: "Q2 allocation review", days_ago: 4 }
  ],
  flags: [{ kind: "regulatory", detail: "SEC filing pending review" }]
}

02 · Healthcare

Clinical decision support, fully cited.

Clinicians and researchers need synthesised answers across peer-reviewed literature, internal hospital protocols, drug interaction databases, and recent case histories — quality-scored so high-confidence sources rank first. Source-level access controls stay intact; PHI never leaves its system, and every claim carries a citation.

// Example query

mantle.semantic_search("treatment-resistant depression combination therapy + cardiac comorbidities", limit: 12)

// What comes back

[
  { score: 0.93, source: "pubmed",   title: "Combination SSRI/SNRI in TRD with cardiac risk",
    journal: "NEJM", year: 2026, design: "meta-analysis" },
  { score: 0.88, source: "internal", title: "Cardiology + Psych protocol v4.2",
    last_reviewed: "2026-03-12" },
  { score: 0.81, source: "ehr_deid", title: "3 closed case studies (de-identified)" }
]
warnings: [{ source: "drug_db", contraindication: "QT prolongation risk" }]

03 · Customer support

Resolve the customer before you respond.

Customer-facing agents fail when CRM ID, ticket ID, and Slack handle are three different identities. Mantle resolves them into a single entity_id so support agents see every open ticket, deal status, and recent escalation across systems before the first reply.

// Example query

mantle.search_entities("Acme Corp", entity_type: "company")

// What comes back

{
  entity_id: "ent_acme_corp_8f2a",
  name: "Acme Corp",
  aliases: ["Acme Corporation", "ACME"],
  sources: ["salesforce", "hubspot", "zendesk", "slack"],
  open_tickets: 3,
  last_escalation: "2026-04-22 · billing dispute",
  csm: "Maren Vosh"
}

04 · Sales enablement

Brief any sales agent in one call.

One call returns the full relationship: emails, calls, contracts, product usage. Drop the result into a sales agent — Claude, Gong copilot, or your own — and it walks into the next meeting fully loaded, with citations.

// Example query

mantle.get_entity_context("ent_acme_corp_8f2a", verbosity: "detailed")

// What comes back

{
  entity: { name: "Acme Corp", arr: "$420k", health: 0.78 },
  recent_activity: [
    { type: "email", subject: "Renewal terms", from: "cto@acme.com", days_ago: 2 },
    { type: "call",  duration_min: 32, summary: "Pricing pushback on enterprise tier", days_ago: 6 },
    { type: "ticket", priority: "P2", status: "open", days_ago: 9 }
  ],
  upcoming_renewal: "2026-07-15"
}

05 · Engineering

Find the canonical doc, fast.

New hires waste days asking which doc is the real one. Mantle finds the latest design doc, current owner, and recent decisions across Notion, Drive, GitHub, and Slack — ranked by recency and quality, not folder structure.

// Example query

mantle.semantic_search("who owns the auth service", limit: 5)

// What comes back

[
  { score: 0.94, source: "notion",  title: "Auth Platform — H1 26 plan",
    owner: "platform-team", updated: "2026-04-19" },
  { score: 0.88, source: "github",  path: "services/auth/README.md" },
  { score: 0.81, source: "drive",   title: "Auth migration RFC v3" },
  { score: 0.74, source: "slack",   channel: "#auth-platform" }
]

06 · Compliance & legal

Search clauses by meaning.

Legal teams need to find the same clause across thousands of contracts and DPAs — phrased differently every time. Mantle indexes by meaning, returns provenance for every hit, and lets you trace back to the source paragraph.

// Example query

mantle.semantic_search("clauses limiting data retention to 12 months", source_filter: ["contracts", "dpa"])

// What comes back

[
  { score: 0.91, source: "drive", file: "Acme MSA 2025.pdf", page: 14,
    excerpt: "...processor shall not retain personal data for longer than 12 months..." },
  { score: 0.87, source: "drive", file: "Globex DPA v3.docx", section: "5.3",
    excerpt: "...12 months from cessation of services..." }
]

07 · Personal

Context across your whole stack.

Ask Claude or Cursor what you committed to last week. Mantle pulls from your Gmail, Drive, Slack, and calendar — across personal and work accounts — read-only, fully revocable, your tokens.

// Example query

mantle.search_entities("Project Aurora") | get_entity_context(...)

// What comes back

{
  entity: { name: "Project Aurora", type: "project" },
  your_commitments: [
    { source: "slack", quote: "I'll have the spec by Friday", date: "2026-04-22" },
    { source: "gmail", quote: "draft posted in Drive", date: "2026-04-25" }
  ],
  next_action: "Spec review — 2026-04-30, 14:00"
}

08 · Analyst & research

Mix structured numbers with unstructured context.

Analyst agents need both — live SQL against your warehouse, plus semantic search across the wikis and docs that explain those numbers. Mantle composes them in one call so the agent reasons over both.

// Example query

mantle.query_database("src_postgres_prod", "SELECT region, SUM(arr) FROM accounts GROUP BY region")
    + mantle.semantic_search("regional pricing strategy 2026")

// What comes back

{
  rows: [
    { region: "EMEA", arr: 4_120_000 },
    { region: "AMER", arr: 6_870_000 }
  ],
  context: [
    { source: "notion", title: "EMEA pricing memo Q2 26", score: 0.89 }
  ]
}