Hormonal Health Research Assistant

The description states that the Hormonal Health Research Assistant is a multi-agent system designed to support endocrine research by separating scientific literature from patient-specific biology using causal graph projection. It uses structured intake, retrieval of scientific literature, and a persistent evidence graph to generate patient-specific projections with biological relevance scoring. The system is built using GPT-5.6 and OpenAI Codex, along with Python.

The project appears to be an early-stage prototype or proof-of-concept submitted for the OpenAI 2026 hackathon. It does not demonstrate any commercial traction, revenue, or customer adoption. The author claims it avoids common AI diagnostic pitfalls by using structured reasoning and pruning irrelevant literature paths, but there is no evidence of real-world usage or performance validation.

The description states that the Hormonal Health Research Assistant is a multi-agent endocrine reasoning system. It includes:

• A 15-agent orchestration framework
• Deterministic intake & biomarker analysis using Pydantic schemas
• Scientific retrieval from PubMed/Semantic Scholar
• Knowledge extraction into structured entities with confidence scores
• Persistent evidence graph stored in SQLite
• Patient-specific graph projection and biological relevance scoring
• Mechanistic explanation generation in Markdown
• A safety layer to enforce non-diagnostic boundaries

It is built using:

• GPT-5.6 (for reasoning and debugging)
• OpenAI Codex (as pair programmer for backend development)
• Python

The system is described as acting as an "evidence-aware reasoning assistant", not an AI doctor.

Inference The product is a research tool, not a diagnostic or clinical decision support system.

The description states that the project was inspired by the limitations of existing AI tools in endocrine disorders. It positions itself as:

• An evidence-aware reasoning assistant, not an AI doctor
• A system that bridges vast scientific literature with individual patient context
• Using structured, uncertainty-aware multi-agent workflows

It claims to avoid two common traps:

1. Generic LLMs producing hallucinated diagnoses
2. Classic RAG systems dumping overwhelming, irrelevant literature

The project also claims to be a multi-agent workflow that separates global knowledge from patient-specific interpretation using causal graph projection.

Inference The positioning is focused on clinical research support and reducing noise in literature-based reasoning, not direct patient care or diagnosis.

The description states that the system targets endocrine and metabolic disorders, such as PCOS and adrenal dysfunction. It is designed for use with patients who have complex, overlapping symptom profiles and conflicting lab patterns.

It is described as a tool for researchers or clinicians working with these conditions, not for direct patient-facing applications.

Inference The ICP likely includes clinical researchers, endocrinologists, or advanced practitioners in reproductive health or metabolic medicine. However, no explicit customer segment or persona is defined.

The description does not provide any information about a business model or pricing structure.

Not evidenced.

The system is built using:

• GPT-5.6
• OpenAI Codex
• Python
• Pydantic schemas
• SQLite database
• Markdown output for explanations

It uses a 15-agent orchestration framework and implements:

• JSON parsing with validation
• State persistence across UI views
• Graph projection and pruning algorithms
• Error handling and API retry logic

The description also mentions migration to native Markdown for user-facing agents and implementation of JSON repair logic.

Inference The system is built in a research or prototyping environment, likely using AI tools for rapid development. It shows technical sophistication but lacks evidence of production-grade delivery or scalability.

The description states that this project was submitted to the OpenAI 2026 hackathon, and no other traction or maturity indicators are provided.

Not evidenced.

The description does not mention any competitors or existing solutions in the market.

Not evidenced.

• The system is described as a research prototype, not a commercial product, raising questions about its intended use and regulatory compliance.
• It is built using GPT-5.6, which may not be available to all users or scalable for production deployment.
• No evidence of real-world testing, validation, or clinical integration.
• The system is described as non-diagnostic but operates in a domain where such clarity is critical.
• The use of OpenAI tools (Codex, GPT) may introduce dependency risks and limit reproducibility.

Inference There is a risk that the project may not be suitable for clinical or commercial deployment without further development and validation.

1. Is this system intended for clinical use, or is it purely for research?
2. What are the regulatory considerations for deploying such a system in healthcare settings?
3. How does the system handle uncertainty in its outputs, and how is that communicated to users?
4. Has the system been tested with real patient data or clinical scenarios?
5. What is the plan for scaling beyond the hackathon prototype?
6. Are there any plans to integrate with existing EHRs or lab systems?