AI Advocate for the Poor

The company appears to be a solo developer project named "AI Advocate for the Poor", submitted as a hackathon entry to OpenAI Build Week 2026. The author describes it as a prototype that uses AI to analyze legal documents and reconstruct case histories, aiming to make institutional processes more transparent and accessible to ordinary people. It is presented as a demonstration of an evidence-first architecture for handling complex procedural information.

• The description states that AI Advocate for the Poor is a demonstration prototype.
• It analyzes prepared Czech legal case studies.
• It reconstructs documents, events, proceedings and contradictions.
• It separates extracted facts, legal interpretation, uncertainty and recommended next steps.
• It supports factual conclusions with exact source quotations.
• It preserves human review instead of presenting AI output as a verdict.
• It processes the demonstrated document workflow locally in the browser.
• The system is designed around four principles: source fidelity, procedural context, explicit uncertainty and human responsibility.

Not evidenced: What specific legal domains or jurisdictions it covers beyond Czech law; what types of documents it processes; whether it supports multiple languages beyond English and Czech; or how it handles document intake or user interaction beyond the prototype's interface.

• The description states that AI Advocate for the Poor aims to "turn complex documents into a transparent, source-grounded map that ordinary people can inspect and challenge."
• It is positioned as a tool to "make justice accessible regardless of wealth, cognitive abilities, health, disability, faith, sexual orientation, or any other personal status."
• The author claims it is a "demonstration prototype, not a lawyer and not a substitute for professional legal advice."
• It is described as a "living proof" that can be extended to support people dealing with public authorities, family and social matters, insurance claims, workplace disputes, consumer problems, journalism and nonprofit counselling.
• The project is framed as a "public value" creation with potential for sustainable services through paid professional workspaces, integrations, document processing and institutional deployments.

Inference: The positioning suggests an intent to democratize access to legal information, but the description does not confirm whether this has been tested or validated in real-world use beyond the prototype.

• The author states that AI Advocate for the Poor is intended for "ordinary people" who face decisions from courts, authorities, insurers, employers and other institutions.
• It targets individuals who "cannot understand the evidence, procedural history or available next steps."
• It also mentions potential users in "lawyers, NGOs, media organizations and institutions", suggesting a possible ICP for professional or institutional use.
• The project is described as supporting people dealing with public authorities, family and social matters, insurance claims, workplace disputes, consumer problems, journalism and nonprofit counselling.

Not evidenced: Specific customer segments beyond the general public; whether there are any identified paying customers or institutional partners; what percentage of users are expected to be individuals vs. professionals.

• The description states that the project can create "public value while developing sustainable services for individuals, lawyers, NGOs, media organizations and institutions."
• It suggests a free public-interest layer supported by paid professional workspaces, integrations, document processing and institutional deployments.
• No pricing information is provided.
• No evidence of revenue streams or monetization strategies beyond the stated model.

Inference: The business model appears to be based on a freemium or tiered service approach, but no details are given about how this would be implemented or whether it has been tested.

• The prototype is built using codex, css, gpt-5.6, html, javascript, openai, pdf, python.
• It processes the demonstrated document workflow locally in the browser.
• It was developed during OpenAI Build Week, with Codex used as an engineering and verification partner.
• The system is designed around four principles: source fidelity, procedural context, explicit uncertainty and human responsibility.
• The current release passes 188 automated tests.
• A live application and source code repository are available.

Not evidenced: Whether the prototype has been scaled or deployed beyond a local browser environment; how it handles large-scale document processing; whether there is any backend infrastructure or cloud integration; or if the system supports real-time updates or collaboration features.

• The project is described as a demonstration prototype, not a product in use.
• It was built during OpenAI Build Week and submitted to a hackathon.
• A live application and source code repository are available.
• The current release passes 188 automated tests.
• Development will continue in a "clearly separated live layer."
• A conditional 2027 field pilot is planned.

Not evidenced: Any user base, customer feedback, or real-world usage data; whether the prototype has been tested with actual users; how many documents it can process or how long it takes to analyze them; or any metrics on performance or scalability.

• The description does not mention specific competitors.
• It is positioned as a tool for legal document analysis and procedural mapping.
• It emphasizes source fidelity, procedural context, explicit uncertainty and human responsibility, which may differentiate it from generic AI tools.
• It is framed as a solution to accessibility issues in legal systems, suggesting a potential overlap with legal tech or AI-assisted legal services.

Not evidenced: Who the direct or indirect competitors are; what existing solutions already exist in this space; how the product compares technically or functionally to those solutions.

• The project is described as a prototype, not a product in use.
• It is explicitly stated that it is not a lawyer and not a substitute for professional legal advice.
• There is no evidence of revenue, customers, or traction beyond the prototype.
• The author states that development will continue in a "clearly separated live layer", but no timeline or funding details are provided.
• The project is built by a single developer (Dušan Dvořák).
• No mention of legal compliance, data privacy, or liability considerations for handling sensitive documents.

Inference: The lack of commercial traction and the single-person team raise concerns about scalability and long-term viability. The emphasis on human review does not address how the system will be maintained or audited at scale.

1. What is the current status of the prototype? Is it being used by any individuals or organizations?
2. How does the system handle document intake and processing beyond the demo case?
3. What are the specific legal or regulatory risks associated with using this tool in real-world scenarios?
4. Are there any plans for monetization beyond the proposed freemium model?
5. How is the system designed to scale beyond a single developer, especially in terms of maintenance and updates?
6. What kind of feedback have you received from users or legal professionals during development?
7. How does the system ensure data privacy and security when processing sensitive documents?