Evidence-Backed Exposure Mapping for SEC Filings

The description states that FragilityGraph is a tool for mapping and simulating financial exposures in SEC filings using AI. It accepts public filing URLs, processes them into chunks, and translates claims into relationship candidates. The system allows users to run scenario simulations that propagate through a graph structure, with results traceable back to source evidence or assumptions. The authors describe building it as an "evidence-backed" tool that avoids inventing missing data.

The project appears to be a proof-of-concept or prototype built for a hackathon, not yet a commercial product. It has no demonstrated traction, revenue, customers or market adoption. The most important open question is whether this concept can scale beyond a hackathon prototype into a viable commercial offering that financial analysts would pay for.

The description states that FragilityGraph:

• Accepts public filing URLs as input
• Returns full documents in bounded chunks
• Translates supported claims into cited relationship candidates
• Builds a trusted graph structure from filings
• Allows running company shock scenarios that propagate through the graph
• Provides traceability of results back to source evidence or declared assumptions
• Features a "review-before-trust" workflow and "evidence desk"
• Uses GPT-5.6 and Codex for development assistance

The system is described as having a FastAPI backend and Next.js frontend, with a financial-terminal-styled interface.

The description states that the authors were inspired by the brittleness of current financial research processes where analysts lose track of material relationships in SEC filings. They aimed to turn filings into a graph that could be "clicked through and stress tested" rather than creating another summarization tool.

The positioning is described as an "evidence-backed AI infrastructure exposure mapping and scenario simulation" tool for SEC filings, targeting financial analysts who need to trace relationships and stress-test companies.

The description states that the tool targets financial analysts who work with SEC filings and need to:

• Find material relationships in 100+ page documents
• Trace exposures when companies are stressed
• Inspect and validate relationships before trusting them
• Run scenario simulations through company graphs

The authors note they built it for "exactly that kind of inspection" but do not specify the exact analyst role or firm type.

Not evidenced. The description does not state any pricing model, monetization strategy, or business model details beyond the self-reported project scope.

The description states:

• Built with FastAPI backend and Next.js frontend
• Uses GPT-5.6 and Codex for development assistance
• Extraction pipeline uses bounded, overlapping chunking instead of naive character cutoff
• Data is not added to trusted graph without human oversight
• Features "review-before-trust" workflow and evidence desk
• Includes scenario propagation animation
• Deployment on Neon, Render, and Vercel
• Uses technologies including: api, codex, css, fastapi, flow, gpt-5.6, neon, next.js, openai, postgresql, pydantic, pymupdf, python, rapidfuzz, react, sqlalchemy, tailwind, typescript, vercel

Not evidenced. The description states this was built for a hackathon and contains no information about revenue, customers, usage metrics, or product maturity beyond the prototype stage.

Not evidenced. The description does not mention any existing competitors or market positioning relative to other financial analysis tools or SEC filing platforms.

• The project is described as a hackathon submission with no demonstrated traction
• No evidence of revenue, customers or commercial adoption
• The authors state they never filled missing values with invented numbers, which may limit practical utility
• The tool appears to be built for a specific use case (SEC filings) rather than broader financial analysis needs
• The "evidence-backed" approach may make it less flexible for exploratory analysis

1. What is the actual market need you're solving for beyond hackathon demonstration?
2. Have you validated this with actual financial analysts or firms?
3. How do you plan to scale beyond the current prototype approach?
4. What are your plans for monetization and pricing?
5. What specific SEC filing types will you support, and how does that map to real-world demand?
6. How do you handle edge cases in document processing and relationship mapping?
7. What is your roadmap for product development beyond this prototype?