A queryable graph built from Form ADV Part 1A across the full RIA and Exempt Reporting Adviser universe — connecting adviser entities, control persons, related persons, private funds, and service providers. Surfaces topology invisible in any single filing: shared control persons, service-provider concentration, and hidden fund families.
Constructs comparable cohorts directly from parsed SEC EDGAR financials, then benchmarks valuation multiples against the cohort distribution rather than a static comp set. Exportable PDF reports with full audit trail back to source filings.
Meaning-based search across registration filings and disclosures. Ask in plain language — "advisers with cross-trade conflicts and affiliated broker-dealers" — and retrieve passages, not keyword noise. Every result links to the underlying filing.
Grades covenant strength and computes the Structural Lender Exposure Index (SLEI) and Credit Agreement Risk Intelligence (CARI) scores across deal structures. This framework is for locating mispricing where covenant quality isn't reflected in spread.
Side-by-side interrogation of private credit benchmarks. CDLI versus academic factor-replication approaches with PME methodology comparisons. Built to answer the allocator's question: what is my direct-lending allocation actually being measured against?
An interactive model of private credit liquidity stress treating redemption cycles as a yield-stress (Bingham plastic) and hysteresis phenomenon. Because flow doesn't begin until stress crosses a threshold, and recovery doesn't retrace the path down.
Seven atomic experiments — one Cloudflare primitive each, one private-markets workflow each — that compose into a miniature fund-administration operating system at the edge.
The simplest private-markets data model that still feels real: LP, fund, commitment, called capital, vintage — with unfunded computed live from SQL.
Upload and retrieve mock GP quarterly reports under fund/quarter keys — the seed of an LP portal document vault: capital call notices, K-1s, DDQ packets.
Ten seeded LPA clauses — key person, GP removal, MFN, recycling — embedded and queried semantically: the core private-markets RAG demo.
Structured inputs — asset class, strategy, target return, risks, client type — into a five-bullet IC-style memo via env.AI.run() on Cloudflare's model catalog.
Submit a capital call notice into a guaranteed-delivery queue; a consumer validates, processes, and writes status back to KV. A real middle-office pipeline in miniature.
One unique object per portfolio session: buy SPY, add a private credit fund, rebalance, reset — mutable state coordinated across requests. An advisor proposal room in embryo.
The "house view" config every demo reads: risk-free rate, private credit spread, default and recovery rates — saved under market-assumptions:base-case, retrieved globally at low latency.
The lab is deliberately atomic — one binding, one workflow, one test per page. But the composition is the point: D1 holds structured fund data, R2 holds documents, Vectorize retrieves meaning, Workers AI explains, Queues process in the background, KV configures, and Durable Objects hold live session state. That is a tiny private-markets operating system hiding inside a weekend experiment.
Every tool on this page is constructed exclusively from public regulatory data — SEC EDGAR full-text and XBRL filings, and Form ADV Part 1A structured data. Pipelines refresh nightly; graph construction and heavy analytics run offline, with precomputed results served from the edge. Nothing here constitutes investment advice; everything here discloses its methodology.
STACK — Python ingestion · PostgreSQL · Neo4j (graph) · Cloudflare Pages (edge delivery) · No tracking, no accounts, no paywall.