Introducing ArcheLearn more

Financial fundamentals you can defend.

Arche is built for teams whose numbers must hold up under scrutiny across models, audits, disclosures and internal review. No silent changes. No unverifiable results.

Request accessView the API documentation

AAPL · Income Statement

Conceptual view of versioned statements (illustrative)

as_of=2023-12-31
Filed2022-02-10Amended2022-04-03Reclassified2023-01-15
Active revisionstatement_version_id: v_7f2a…

Revenue

Change captured as a first-class event (not an overwrite)

from382.47Bto383.28B10-K/A
Provenance attachedTemporal truth explicitAs-of queries supported
Why Arche exists

Because financial systems erase accountability.

When filings change, most data platforms overwrite history. You can’t prove what changed, when it changed or what decisions relied on the prior numbers. Arche fixes this by treating change as auditable data.

SourcesRaw filings and corrections become versioned eventsEDGAR filingsAccessioned, timestampedAmendments / restatementsNew versions, never overwritesProvider correctionsExplicit change eventsVersioned ledgerImmutable versions with provenanceStatement: AAPL • 2024-12-31source=edgar • accession=0000320193-25-000010version=v3 • ingested_at=2025-02-05T02:14:31Zprovenance: strongVersions over timev1v2v3Contract-first APIVersioned routes, machine-readable schemasGET /v1/…Deterministic orderingExplicit paginationCanonical error envelopesStable schema contractsReady for MCP tool usemcp
Trust, by design

Built to be consumed by systems.

Arche exposes financial data through stable, versioned contracts designed for direct consumption by software, not interpretation. The same guarantees that preserve historical truth make integration predictable and safe.

AI- and agent-compatible.
Arche exposes machine-readable, deterministic tools designed for direct use by AI agents via the Model Context Protocol (MCP), without scraping or inference.
Versioned financial history.
Corporate fundamentals evolve. Arche stores statements as immutable versions, so you can reproduce what was known at any point in time—without reconstructing the past.
No silent overwrites.
Corrections and restatements never replace historical data. Changes are explicit, traceable and additive—preserving reproducibility across research and models.
Provenance by default.
Every value is traceable to its source filing and ingestion event. That makes results explainable, reviewable and defensible when scrutiny matters.
Deterministic retrieval.
Identical queries return identical results. Stable ordering, explicit pagination and predictable semantics are enforced by design—not convention.
Contract-first API.
A versioned HTTP API with explicit schemas and clear error semantics. Built to integrate cleanly into internal systems, pipelines and analytical workflows.
The Arche difference

What makes Arche different

Built for quant teams, valuation workflows and AI research systems that need data they can trust.

Feature comparison

Fancy Terminals

Capabilities
Point-in-time queries
Yes
Restatement lineage
Yes
Accounting reconciliation
Manual
MCP/AI-native interface
No
Full audit trail
Yes
Valuation-grade precision
Yes

Arche

Capabilities
Point-in-time queries
Yes
Restatement lineage
Yes
Accounting reconciliation
Automatic
MCP/AI-native interface
Yes
Full audit trail
Yes
Valuation-grade precision
Yes (Decimal 38,6)

Consumer APIs

Capabilities
Point-in-time queries
No
Restatement lineage
Partial
Accounting reconciliation
No
MCP/AI-native interface
No
Full audit trail
No
Valuation-grade precision
Varies
Capabilities

Questions you can answer precisely

Arche is designed around the questions financial systems actually need to answer, without inference, reconstruction or guesswork.

Point-in-time truth
What did we know on Dec. 15, 2024? Arche resolves immutable, point-in-time snapshots using accepted filing dates. No look-ahead bias or retroactive changes. Every result is reproducible.
Restatement impact
What changed, and how material was it? Every restatement is preserved and comparable. Arche computes per-metric deltas and classifies materiality so you know exactly what changed and whether it mattered.
Accounting correctness
Do these statements actually reconcile? Arche verifies balance sheet equations, cash-flow rollforwards and segment totals. Each check produces PASS / WARNING / FAIL with exact expected vs. actual values.
Model-ready metrics
Can this be used directly in models? Arche exposes pre-computed valuation and performance metrics with Decimal precision. When a metric can’t be computed, the reason is explicit, never a silent null.
Data quality overlay
Is this number plausible? Every normalized fact is checked for presence, sign validity and historical outliers. Anomalies include severity and explainable baselines — not opaque flags.
Agent-safe interface
Can agents reason over this safely? Arche exposes deterministic, schema-validated tools via the Model Context Protocol (MCP), allowing AI agents to query, reconcile and analyze financials without scraping or inference.