PlatformFor RIAsIntegrationsSecurity

Article — The Platform

Four pillars,
one document brain.

DocumentBoost is one platform with three frontends — a marketing site, an advisor back-office, and a white-label client portal — all built on top of the same document and profile layer.

I.

Document Brain

Every document a firm owns, ingested, classified, extracted, embedded, citation-linked.

  • § 01Bulk ingest via email forwarding, batch import, and integrations
  • § 02LLM-driven document classifier — sub-docs, K-1s, trust docs, KYC, account opening
  • § 03Semantic search (pgvector) plus keyword search across the firm's entire estate
  • § 04Per-firm form library, pre-loaded during white-glove onboarding
II.

Profile Layer

Households, people, entities, and attributes — with provenance, history, and citations.

  • § 01Household model first-class, not bolted on top of contact lists
  • § 02Attribute history — every value sourced to the document that produced it
  • § 03Multi-entity support for trusts, LLCs, and account structures
  • § 04Profile reconciliation across CRM, custodian, and performance platforms
III.

Workflow & AI

Guided Fill, send-to-sign, tasks, messaging, compliance — anchored to documents and profiles.

  • § 01Guided Fill copilot for sub-docs and account-opening packets
  • § 02Send-to-sign via Docusign with embedded signing in the client portal
  • § 03Task queue, messaging threads, and AI-drafted advisor replies
  • § 04Compliance scanning, audit export, and ADV/CRS delivery tracking
IV.

Client Experience

White-label portal for households, spouses, and external collaborators.

  • § 01Per-shop branding and custom domains — clients see your firm, not ours
  • § 02Action queue, embedded signing, and transparent document view
  • § 03Multi-party access — primary, spouse, CPA — with scoped permissions
  • § 04Messaging on the same surface clients already use

Article V — Recent Capability Surfaces

V.01

Workflows

Maestro-style automation, but for wealth-management back-office. Declarative trigger → conditions → actions. No code, no consult fees — see ADR-036.

V.02

MCP — Claude & ChatGPT

Your existing Claude or ChatGPT can query your firm's DocumentBoost data directly via bearer-token MCP — see ADR-035 and the developers section for setup.

V.03

Type templates & extraction

Define a document type once — ADV, K-1, subscription agreement — and extract attributes from every future filing the same way. DocuSign-Navigator-inspired structure.

V.04

Trust posture

67 trust-boundary tests on consent, anti-CSRF, startup invariants, and notification fan-out — plus an immutable WORM-enforced audit log. See the security page.

V.05

Matrix — documents × questions

Pick a stack of documents, pose the same questions of every one, and read the answers in a single grid. Cells are cited back to the chunk, and a confidence pill surfaces the retrieval signal so the advisor can tell a high-grounded answer from a speculative one.

  • § 5.01Rows = documents, columns = questions, cells = cited answers. Built on the same retrieval primitives as Ask AI — every cell rolls up into the same AI Brain trace.
  • § 5.02Per-cell side panel for citations, confidence, and one-click recompute. Failed cells retry inline.
  • § 5.03Useful for diligence on a fund-document stack, comparing ADV filings across managers, or pulling the same field out of every K-1 in a household.

The full platform brief — architecture, integrations, and roadmap — is available on request to qualified firms.

Request the Brief

Authorized Signature