Archive position — measured, not model output
1 like on Devpost
506 of the 7,856 archived projects have more likes, and 1,758 share exactly 1 — so this project's #1,056 place in the like-ranked listing is a tie-break inside that group, not a ranking.
Projects (log scale)
Likes on Devpost. ▲ marks this project's group.
Show the figures
| Likes | Projects | Share of archive |
|---|---|---|
| 0 | 5,592 | 71.2% |
| 1 | 1,758 | 22.4% |
| 2 | 285 | 3.6% |
| 3–4 | 132 | 1.7% |
| 5–9 | 75 | 1.0% |
| 10+ | 14 | 0.2% |
Executive Summary
What the company appears to be
Fibber is an adversarial clinical reasoning platform designed for medical education. It uses a large language model (GPT-5.6) to simulate difficult patients who lie, minimize symptoms, conceal information, or misdirect during simulated patient interviews. The system is built around a deterministic engine that enforces rules and scoring, while the LLM plays the role of a deceptive patient.
What changed
The project evolved from an idea to build a tool for bedside teaching in emergency medicine into a platform that simulates adversarial clinical interactions using AI. It was submitted as part of the OpenAI 2026 hackathon.
Single most important open question
Does this platform have any evidence of traction, revenue, or adoption by medical institutions? The description is entirely self-reported and lacks any data on usage, customers, or commercial viability beyond the author's own account.
What The Product Actually Is
The description states that Fibber is an adversarial clinical reasoning game. It simulates difficult patients using GPT-5.6 who follow deception patterns (minimizer, concealer, misdirector). The system includes:
- A deterioration clock tied to scripted vitals
- Question caps per encounter
- Simulated stewardship budget
- CTF-style flag scoring for extracted truths
- Socratic examiner that rebuilds reasoning after diagnosis
- Deterministic fallback for offline judging
The platform uses a rule-based engine where the model may act but never decides what is true. Clinical truth lives in authored JSON cases validated by Zod at startup.
Positioning & Claim Evolution
The author claims Fibber addresses diagnostic errors in medicine, which they state kill 371,000 people annually in the US. They position it as a solution to the problem that standardized patient actors are expensive and hard to scale, especially since the USMLE retired Step 2 CS in 2021.
The platform is positioned as an alternative to traditional medical education methods that don't simulate real patient behavior. The author states they want to pilot Fibber with their own students against standard vignette teaching to measure whether flag extraction transfers to standardized patient performance.
Target Customer & ICP
The description states the target customer is medical students and clinicians in training, particularly those in emergency departments. The author works in an emergency department and trains medical students at the bedside.
The platform appears designed for:
- Medical educators
- Emergency medicine trainees
- Institutions seeking scalable clinical reasoning training
However, there's no evidence of specific institutional adoption or customer data.
Business Model & Pricing Evidence
Not evidenced. The description does not contain any information about pricing, revenue streams, or business model.
Technical & Delivery Signals
The platform is built with:
- GPT-5.6 (called through official Responses API)
- TypeScript engine
- React interface
- Express API with rate limits and session serialization
- Codex for schema, engine, API, and test suite
- JSON-based case validation with Zod
- Deterministic fallback for offline use
The system includes:
- Rule enforcement via deterministic engine
- Fact-citation protocol with server-side rejection
- 48 automated tests for game balance
- Configurable model ID (defaults to gpt-5.6-sol)
- Schema validation and test suite built by Codex
Traction & Maturity Signals
Not evidenced. The description contains no information about:
- Revenue or funding
- Customer base or adoption
- Usage metrics
- Product maturity beyond prototype stage
- Market traction
The author mentions live testing with their own students, but this is not quantified.
Competitive Context
Not evidenced. The description does not contain any information about competitors, market positioning, or competitive landscape.
Key Risks & Red Flags
Key risks
- No evidence of commercial viability or traction
- Platform appears to be a prototype built for a hackathon
- Reliance on GPT-5.6 which may not be available at scale
- Limited to one team member (Rifqi Haikal)
- No evidence of institutional adoption or market validation
Red flags
- Self-reported only, no independent verification
- No revenue, customer, or traction data provided
- Platform built for a single developer with no apparent commercialization plan
- Limited to three case examples (inferior STEMI, occult GI bleed, spinal epidural abscess)
- No evidence of scalability beyond prototype
Diligence Questions To Ask The Founders
- What is the actual market need for this platform? How many medical schools or training institutions are interested in adopting it?
- Have you conducted any pilot studies with actual students or educators?
- What are your plans for scaling beyond the current prototype?
- How do you plan to monetize this platform?
- What specific feedback have you received from medical educators about the platform's utility?
- Are there any institutional partnerships or pilot programs in progress?
- What is the timeline for moving from prototype to commercial product?
- How do you plan to handle regulatory and ethical considerations around AI in medical education?
Investment/Partnership Verdict
Not evidenced. The description provides no information about:
- Financial performance
- Customer base or adoption
- Revenue streams
- Market traction
- Commercial viability
- Funding status
The platform appears to be a prototype built for a hackathon by a single developer (Rifqi Haikal). There is no evidence of commercialization, institutional adoption, or market validation beyond the author's own account. The description states that this is an unverified self-report and lacks any data on revenue, customers, or traction.
Source
Submitted to the OpenAI 2026 hackathon on Devpost. Project home on DevPost.
The analysis above was generated by a language model from the project's own one-line description. It is not independent research and contains no verified traction, revenue or customer data.