Threshold86

What the company appears to be: Threshold86 is a self-reported software-delivery platform designed for highly regulated, disconnected environments. It claims to automate the process of converting approved requirements into tested, auditable, offline-ready software releases using AI (specifically GPT-5.6 and Codex). The system is described as running locally, with no external API calls post-installation.

What changed: The project was built in less than 15 hours by a single developer as part of an OpenAI hackathon submission. It demonstrates a minimal viable product (MVP) that includes automated testing, security scanning, traceability, and offline packaging for a fictional manufacturing system.

Single most important open question: Is there any evidence that Threshold86 has been adopted or tested in real-world environments beyond the hackathon MVP?

Analysis basis: This report is based entirely on the self-reported description provided by the author. No external verification, traction data, revenue figures, customer names or third-party sources are available.

• The description states that Threshold86 is a "governed software-delivery and evidence layer."
• It uses GPT-5.6 and Codex to convert requirements into structured engineering work.
• The system generates automated tests, security scans, dependency audits, database migrations, traceability, and offline-ready release packages.
• It claims to verify:
  • Automated tests
  • Code coverage (99.76%)
  • Type checking (mypy)
  • Security scans (Bandit)
  • Dependency audits (pip-audit)
  • Database migrations and rollback behavior
  • Requirement-to-test traceability
  • SBOM generation
  • SHA-256 manifest for release artifacts
  • Tamper-detection tests
  • Offline runtime independence
• The MVP runs locally using Python, FastAPI, SQLite, Alembic, pytest, and GitHub Actions.
• It is built to operate without internet or OpenAI API calls after installation.

Inference: The product appears to be an AI-assisted development pipeline tailored for compliance-heavy environments where software must be validated before crossing controlled boundaries. However, this is based on the author's own account and not independently verified.

• The tagline "Nothing crosses without evidence" positions Threshold86 as a compliance-enabling tool.
• The description frames it as a solution for regulated environments where software changes require more than just working code — they must also provide verifiable evidence of testing, review, traceability, and reproducibility.
• The author claims that AI (GPT-5.6/Codex) was used throughout the full engineering lifecycle: from requirements to implementation, migrations, tests, security checks, documentation, debugging, and demo prep.
• It is positioned as a tool for preparing software for formal review, not replacing independent assessments or Authorizing Officials.

Inference: The positioning suggests a niche market focused on enterprise compliance and controlled environments. The claim evolution shows a shift from AI-assisted development to a governance-focused delivery system.

• The description implies that Threshold86 targets "controlled, disconnected, and highly regulated corporate environments."
• It is designed for situations where software must be tested, reviewed, traceable, reproducible, and ready to move across a controlled boundary.
• The fictional target application (ForgeLine MES) is described as a manufacturing execution system, suggesting potential use in industrial or manufacturing contexts.

Not evidenced: No specific customer segments, industries, or personas are named. No evidence of actual customers or pilot programs beyond the hackathon MVP.

• The description does not mention any pricing model, subscription tiers, or monetization strategy.
• There is no indication of whether Threshold86 will be sold as a SaaS offering, on-premises software, or open-source tool.
• No revenue streams, licensing terms, or commercial partnerships are mentioned.

Not evidenced: No business model or pricing information is provided in the description.

• The system is built using Python (FastAPI, SQLModel, Alembic), TypeScript (React), Docker, GitHub Actions.
• It supports offline-first operation with zero external API calls post-installation.
• Key technical features include:
  • Automated testing (pytest)
  • Code coverage (99.76%)
  • Type checking (mypy)
  • Security scanning (Bandit)
  • Dependency auditing (pip-audit)
  • Database migrations and rollback behavior
  • Requirement-to-test traceability
  • SBOM generation
  • SHA-256 manifest for release artifacts
  • Tamper-detection tests
• The system is claimed to be reproducible via GitHub Actions workflow.

Inference: Technical signals suggest a strong focus on automation, compliance, and offline readiness. However, these are self-reported and not independently validated.

• The MVP was completed in less than 15 hours by one developer.
• It includes:
  • 111 automated tests
  • 99.76% code coverage
  • Strict mypy validation
  • Ruff formatting and linting
  • Bandit security scanning
  • Dependency auditing with pip-audit
  • Tested Alembic upgrade and rollback behavior
  • Requirement-to-test traceability
  • SBOM generation
  • SHA-256 manifest covering 47 release artifacts
  • Tamper-detection tests
  • Verified offline Version 1.0-to-Version 1.1 upgrade
  • Zero measured OpenAI or external API calls during disconnected operation

Not evidenced: No evidence of customer adoption, usage metrics, or real-world deployment beyond the hackathon MVP.

• The description does not name direct competitors.
• It implies a niche market for regulated environments where software must be validated before crossing boundaries.
• Competitors in this space might include:
  • Software delivery platforms with compliance features (e.g., those used in aerospace, defense, or healthcare)
  • DevOps tools that support audit trails and traceability
  • AI-assisted development platforms focused on code quality and security

Not evidenced: No competitive analysis or market positioning beyond the author’s own claims.

• The entire project was built in less than 15 hours by a single person.
• There is no evidence of real-world testing, customer feedback, or production use.
• The system relies heavily on AI (GPT-5.6/Codex) for development, which may introduce risks around accuracy, control, and scalability.
• The author explicitly states that the AI supports the engineering process but does not replace human responsibility for scope, architecture, security boundaries, and release decisions — this could be a risk if not properly implemented in practice.
• No mention of enterprise integration capabilities or scalability beyond the MVP.

Inference: High risk due to lack of real-world validation, reliance on AI for core functions, and limited development time.

1. What specific regulated environments have you tested Threshold86 in?
2. How does the system handle scope creep when using AI-assisted development?
3. Can you demonstrate how the system enforces mandatory failures during release validation?
4. Is there any plan to integrate with existing enterprise AI platforms or compliance frameworks (e.g., OSCAL)?
5. What are your plans for scaling beyond the MVP and ensuring long-term maintainability?
6. How do you ensure that human oversight remains effective in an AI-augmented workflow?