OpenAI 2026 hackathon

PACT — Proof-Carrying Autonomous Change

PACT autonomously fixes software bugs using evidence-driven AI: it analyzes failures, generates and applies patches, verifies the result, and automatically rolls back unsafe changes.

Solo project by Jayasurya Jayasurya · 0 likes · 0 comments

Archive position — measured, not model output

0 likes on Devpost

2,264 of the 7,856 archived projects have more likes, and 5,592 share exactly 0 — so this project's #5,796 place in the like-ranked listing is a tie-break inside that group, not a ranking.

Projects (log scale)

1
10
100
1k
10k
05,592
11,758
2285
3–4132
5–975
10+14

Likes on Devpost. ▲ marks this project's group.

Show the figures
LikesProjectsShare of archive
05,59271.2%
11,75822.4%
22853.6%
3–41321.7%
5–9751.0%
10+140.2%
Devpost like counts for all 7,856 archived projects, captured when this archive was built.

Executive Summary

What the company appears to be

PACT is described as a system that autonomously fixes software bugs using AI-driven reasoning and verification. It claims to analyze failures, generate patches, apply them, verify results, and roll back unsafe changes — all without human intervention.

What changed

This project was submitted to the OpenAI 2026 hackathon, indicating it is a prototype or proof-of-concept built in a short timeframe. The description does not indicate any prior development, traction or commercialization beyond this submission.

Single most important open question

Is there evidence of a real-world use case or customer need that justifies further development or investment? The description provides no indication of adoption, revenue, or even whether the system works as claimed.

Back to contents

What The Product Actually Is

The description states:

"PACT autonomously fixes software bugs using evidence-driven AI: it analyzes failures, generates and applies patches, verifies the result, and automatically rolls back unsafe changes."

This implies a tool that uses AI to identify and resolve software defects. It is positioned as an autonomous system that operates end-to-end — from failure detection to patch application and rollback.

Evidence

  • The author describes PACT as a system that "analyzes failures", "generates and applies patches", "verifies the result", and "automatically rolls back unsafe changes".
  • It is built with technologies like FastAPI, GPT (version 5.6), Python, React, SQLite, and OpenAI API — suggesting it's a software tool using AI for code repair.

Inference It is likely a prototype or hackathon project, not a production-ready product.

Back to contents

Positioning & Claim Evolution

The tagline states:

"PACT autonomously fixes software bugs using evidence-driven AI: it analyzes failures, generates and applies patches, verifies the result, and automatically rolls back unsafe changes."

Claim

PACT is an autonomous system that uses AI to fix bugs without human intervention.

Evidence

  • The description claims PACT is “autonomous”.
  • It positions itself as using “evidence-driven AI” — implying reasoning or logic-based approaches rather than just pattern-matching.
  • It includes a rollback mechanism, suggesting it is designed for safety in production environments.

Inference The claim is ambitious but unproven. The system appears to be a conceptual framework, not a tested solution.

Back to contents

Target Customer & ICP

Evidence

  • No explicit customer or persona described.
  • The project is presented as a hackathon submission with no mention of target users or industries.

Inference The target audience is likely software engineers or DevOps teams who work on bug fixing and code maintenance. However, this is speculative — the description does not confirm it.

Back to contents

Business Model & Pricing Evidence

Evidence

  • No pricing model, licensing terms, or monetization strategy described.
  • No indication of whether PACT is intended for internal use, SaaS delivery, or open-source distribution.

Inference The business model is unclear. It may be a prototype with no commercial intent at this stage.

Back to contents

Technical & Delivery Signals

Evidence

  • Built with: FastAPI, GPT (5.6), JavaScript, OpenAI API, Pydantic, pytest, Python, React, SQLAlchemy, SQLite, Uvicorn, Vite.
  • The system is described as “autonomous” and uses AI to generate and verify patches.

Inference The tech stack suggests a full-stack prototype with backend (FastAPI, Python), frontend (React), and AI integration (OpenAI API). It likely runs locally or in a sandboxed environment.

Back to contents

Traction & Maturity Signals

Evidence

  • Submitted to the OpenAI 2026 hackathon.
  • Team size: 1 person (Jayasurya Jayasurya).
  • No mention of customers, users, revenue, or product adoption.
  • No evidence of prior funding, partnerships, or product releases.

Inference This is a very early-stage project — likely a prototype or proof-of-concept. There is no evidence of traction or maturity.

Back to contents

Competitive Context

Evidence

  • No mention of competitors or existing solutions in the space.
  • The description does not reference similar tools or platforms for automated bug fixing or AI-assisted code repair.

Inference It’s unclear whether this project addresses a known market gap, overlaps with existing tools, or introduces something novel. No competitive landscape is evident.

Back to contents

Key Risks & Red Flags

  1. Unproven claims: The system is described as autonomous and AI-driven but lacks evidence of functionality.
  2. No traction or adoption: No customers, revenue, or usage data are provided.
  3. Single founder: A team size of one may limit execution capacity.
  4. Hackathon prototype: The project was submitted to a hackathon — not a commercial product.
  5. Lack of validation: There is no evidence that the system actually works as described.

Back to contents

Diligence Questions To Ask The Founders

  1. What specific bugs or failure scenarios does PACT handle?
  2. How does it verify that patches are correct and safe?
  3. Has it been tested on real-world codebases or in production environments?
  4. What is the intended deployment model (e.g., SaaS, on-prem, CLI)?
  5. Are there any existing users or partners interested in using this tool?

Back to contents

Investment/Partnership Verdict

Verdict Not evidenced.

Confidence Low. The description provides no evidence of traction, revenue, customer adoption, or even a working prototype. It is a self-reported hackathon submission with no commercial or technical validation.

Inference This project is at an extremely early stage and does not yet demonstrate a viable product or market need. Any investment or partnership would be highly speculative without further evidence of functionality, traction, or customer interest.

Back to contents

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.