Shipped through v0.15.7ABS-first workflowReview + model + API

From prospectus ingestion to review, tie-out, and model delivery

Boxfish is no longer just a document-to-model demo. The current 0.15 release line covers extraction, review and verify workflows, completeness gates, model drill-down, servicer certificate ingestion, public APIs, and audit-visible delivery paths.

AI extraction with recovery paths
Readiness gating before handoff
API and Python SDK delivery

Join the waitlist for early access. Product positioning below is based on shipped Boxfish functionality and public competitor pages.

Current workflow scope
Boxfish v15 capability map
Live
Prospectus extraction
Deal structure, collateral, fees, glossary, date interpretation
Review and Verify
Boxing overlays, comments, flags, assignments, audit trail
Model generation
Excel export, drill-down, tie-out, dec validation, WAL
Readiness and certificates
Coverage gates, readiness reports, signing visibility, imports
Public delivery layer
v1 API, Python SDK, comparison, anomaly, glossary access
End-to-end coverage
One deal. Every step.
Extraction, review, model generation, tie-out, readiness gating, and delivery all verified on a real ABS deal from ingestion to signed output.
Minutes
From PDF to structured deal data
A 200+ page prospectus goes through extraction, tranche mapping, waterfall rules, and collateral data in a single workflow — not a multi-week manual pass.
Formula-driven
Excel models, not static spreadsheets
Every exported workbook carries live formulas across scenario, tranche, and waterfall sheets. Analysts can stress-test assumptions without rebuilding the model from scratch.
End-to-end
Extraction to tie-out in one surface
Review, flagging, model generation, dec table validation, and completeness gating all run in the same system — no re-keying data between disconnected tools.
Eliminated
Manual prospectus rebuilds
The work that used to cost hundreds of thousands of dollars per deal in analyst hours — reading, extracting, verifying, and modeling — runs as a structured, auditable workflow.
Workflow

One operating surface instead of five disconnected steps

The landing page used to describe Boxfish like a narrow model generator. The product now spans ingestion, review, QA, modeling, and delivery, so the story needs to match the actual system.

01
Ingest the deal package

Upload the prospectus, trigger extraction, and let Boxfish catalog deal structure, collateral, fees, and glossary terms.

02
Review and verify in context

Work in a structured review workspace with PDF overlays, comments, flags, assignments, and source-linked corrections.

03
Generate and inspect the model

Build a formula-driven Excel model, open drill-down views, validate dec tables, and compare results against the document.

04
Gate, certify, and integrate

Run readiness checks, expose outputs via API and SDK, and move into certificates and downstream reporting workflows.

Shipped in v15

The cool features are no longer hypothetical

Everything below is pulled from shipped 0.15.0 through 0.15.5 changes, not a future roadmap. This is the current Boxfish surface area that the public page should sell.

Extraction Intelligence
Prospectus extraction with issuer-aware recovery paths
The 0.15 line is not just OCR plus a spreadsheet. It includes structured extraction, glossary persistence, temporal date validation, and confidence-aware recovery paths.
  • Deal info, tranches, waterfall rules, collateral, fees, glossary terms
  • Temporal date engine for vague date qualifiers and contextual resolution
  • Issuer-specific prompt addenda plus deterministic fallback and CANNOT_VERIFY capture
Review Workspace
Review and Verify with overlays, flags, comments, and assignments
Boxfish now covers the human review layer directly: navigation, boxing overlays, flags, threaded comments, section assignments, diffing, and guest review flows.
  • Three-panel boxing workspace with PDF overlay rendering
  • Comments, reactions, assignments, certification timeline, and audit trail
  • Readiness report panel and assertion extraction CTA for empty review pages
Modeling
Model generation with drill-down and persisted projections
Version 15 added CalcAgent to model UI integration, projection persistence, and period-level drill-down so users can inspect what the waterfall engine actually did.
  • Excel model generation with full formula logic and scenario sheets
  • CalcAgent drill-down from model outputs to period-level waterfall detail
  • WAL, dec table validation, tie-out, and projection persistence for reuse
Readiness and QA
Completeness gates before handoff
The product now blocks weak handoffs with configurable completeness thresholds and exposes coverage, confidence, and missing-assertion gaps before outputs move downstream.
  • Pre-handoff completeness gate with threshold enforcement
  • Coverage dashboards, readiness reports, and low-confidence gap surfacing
  • False-positive validation fixes and production hardening in 0.15.3 to 0.15.5
API and Delivery
Public API surface and Python SDK for system integration
The 0.15 branch expanded Boxfish from a UI-only workflow into an integration layer with public API endpoints, OpenAPI coverage, and SDK methods.
  • Deal comparison, anomaly detection, glossary, tie-out, and dec validation endpoints
  • Python SDK coverage for the v1 surface area
  • API-key access patterns for enterprise and partner workflows
Certification and Ops
Servicer certificate ingestion and signed output visibility
Boxfish now handles servicer certificate imports, HMAC signing enforcement, signed-status visibility, and health reporting around certificate readiness.
  • CSV and Excel servicer certificate import with format detection
  • Certificate HMAC enforcement and signing-status exposure
  • Health and operational visibility for production delivery paths
Competitive Landscape

Research-backed positioning against the current market

Based on publicly available product pages, the market leaders around structured finance still cluster around model libraries, surveillance, or loan-level data infrastructure. Boxfish is differentiated by covering the prospectus-to-review layer that usually gets stitched together manually.

Comparison area
Boxfish
Intex
dv01
Trepp
Publicly marketed center of gravity
Prospectus-to-review-to-model workflow
Structured finance model library and analytics
Loan-level data infrastructure and reporting
CMBS/CLO surveillance, research, and valuations
Data and analytics emphasis
Document extraction, tie-out, review gating, and workflow traceability
Cash flow models, deal library, APIs, and Excel plug-ins
Validated loan-level data, deal reporting, cash flows, and data delivery
Deal, loan, and property intelligence with generated cashflows and alerts
Why it matters for teams
Replaces the manual handoff between document reading, verification, and model output
Deep modeling depth once the deal is already structured in the platform
Strong foundation when the core problem is collateral data normalization and reporting
Strong fit when the workflow is surveillance, valuation, and market monitoring
Intex
Public positioning emphasizes prebuilt structured finance models, a large deal library, APIs, and Excel plug-ins.
dv01
Public positioning emphasizes validated loan-level data, deal reporting, cash flows, and AI-ready delivery into Snowflake, BigQuery, S3, and related stacks.
Trepp
Public positioning emphasizes CMBS and CLO databases, generated cashflows, surveillance, daily valuations, and event-driven market monitoring.

Research note: the matrix reflects publicly marketed focus areas on official product pages accessed April 4, 2026. It is a positioning summary, not a claim that competitors cannot support adjacent workflows through services, custom work, or non-public features.

For analysts
Replace manual document mining and spreadsheet stitching with a focused review and modeling workspace.
  • AI-assisted extraction instead of manual document mining
  • A structured review workspace instead of side spreadsheets and screenshots
  • Model generation and tie-out in the same system as verification
For enterprise teams
Standardize review, handoff, and delivery paths without losing audit visibility or downstream integration options.
  • Readiness gates before handoff and certification
  • Audit-ready timelines, section assignments, and guest review flows
  • API and SDK delivery for integration into existing operating models
Why it matters

Move from manual review weeks to a controlled AI workflow

Boxfish compresses document extraction, review, tie-out, and delivery into one operating surface so teams can spend more time on decisions and less time rebuilding the same deal context in multiple tools.

Minutes, not weeks
Cut the mechanical work between upload, review, model generation, and export.
Review with accountability
Assignments, comments, audit history, and readiness gates stay attached to the deal.
Catch issues earlier
Tie-out, dec validation, anomaly detection, and completeness metrics surface gaps before handoff.
Deliver the result
Push outputs into Excel, certificates, and API consumers without rebuilding the workflow.
Proof strip
Review workspace with overlays, flags, comments, assignments, and audit trail
Completeness gate plus readiness reporting before outputs move downstream
Public API and Python SDK for comparison, anomalies, glossary, tie-out, and validation
Servicer certificate import plus signing-status visibility and operational checks

Join the Boxfish waitlist

If you want the current v15 workflow on your own deals, join the list. We will reach out when access opens and when we are ready to onboard design partners.

Boxfish
AI securitization workflow spanning prospectus extraction, review, modeling, tie-out, and delivery.

Disclaimer: AI-generated outputs are for workflow acceleration and review support. Users remain responsible for validating results against official prospectus and support documents.