Expediente Cero

The description states that Expediente Cero is a human-centered workspace for administrative case handling, using AI to prepare synthetic cases for professional review. It supports three procedures: self-employment registration, employee hiring, and grant requests.

It uses:

• GPT-5.6 via OpenAI's Responses API with Structured Outputs
• Python (FastAPI) backend with SQLAlchemy, Alembic, SQLite
• Next.js/React frontend with TypeScript
• GitHub Actions for CI/CD

The system:

• Converts input messages into structured facts using GPT-5.6
• Applies rule-based validation independently of the model
• Allows human reviewers to edit and approve/reject cases
• Maintains an audit timeline of all actions

Not evidenced No evidence of actual customers, revenue, or live deployment beyond the prototype.

The description states that Expediente Cero is:

• A human-centered workspace
• Designed to avoid AI inventing facts, applying hidden rules, or making final decisions
• Built to structure administrative cases and support professional review workflows
• Focused on controlled AI use, where the model prepares information but does not decide

It positions itself as a tool that:

• Uses AI for data extraction and drafting
• Keeps rule-based validation separate from AI outputs
• Ensures human authority remains in final decisions

Inference The positioning implies a focus on compliance, safety, and auditability in regulated or formal administrative environments.

The description states that Expediente Cero supports administrative professionals handling:

• Self-employment registration
• Employee hiring
• Grant requests

It is designed for users who must:

• Organize incomplete information
• Identify missing facts
• Contact applicants
• Document decisions

It supports Spanish and Galician languages.

Not evidenced No evidence of actual target customers, user personas, or market research beyond the prototype's scope.

The description states that Expediente Cero:

• Uses only synthetic test data
• Does not submit applications
• Does not determine eligibility
• Does not replace professional advice

It is described as a prototype, not a commercial product.

Not evidenced No pricing model, monetization strategy, or revenue streams are described. No evidence of paid users or subscriptions.

The project was built with:

• Backend: Python (FastAPI), SQLAlchemy, Alembic, SQLite
• Frontend: Next.js, React, TypeScript
• AI integration: GPT-5.6 via OpenAI Responses API with Structured Outputs
• CI/CD: GitHub Actions
• Deployment: Render

It uses:

• Typed schemas throughout the API and domain layers
• Independent validation of model outputs
• Immutable history (original model output preserved, edits stored as versions)
• Strict schema enforcement to prevent silent guessing of missing data

Inference The architecture suggests a focus on safety, traceability, and separation of concerns — key features for regulated workflows.

The description states:

• This is a hackathon submission
• Built in one week
• Uses only synthetic test data
• Does not submit applications or replace professional advice
• No real-world deployment or customer feedback

Not evidenced No evidence of traction, adoption, or usage beyond the prototype.

The description does not mention any competitors. It is a self-contained project with no reference to existing tools in the administrative case management or AI workflow space.

Not evidenced No competitive analysis, market positioning, or differentiation from other tools.

• Prototype-only: The product is described as a hackathon submission with no commercial traction.
• No real-world use: It uses only synthetic data and does not submit applications or replace professional advice.
• Unproven business model: No pricing, monetization, or customer validation.
• Limited scope: Only three procedures are supported; no indication of scalability or extensibility.
• AI dependency: Relies heavily on GPT-5.6, which may not be available or stable in production.

1. What is the intended transition from prototype to product?
2. Are there any real-world use cases or pilot programs planned?
3. How does the team plan to validate business requirements and user needs beyond the prototype?
4. Is there a plan for secure authentication, role-based access, or encrypted document storage?
5. What are the long-term plans for AI model availability and cost management?