ChangeProof

The description states that ChangeProof is a system that:

• Converts an external obligation (e.g., a regulatory change) into a cited, reviewable, and tested organizational change package.
• Extracts the exact source quote and computes affected workspace spans with evidence for every graph edge.
• Proposes coordinated changes across policies, procedures, controls, application logic, schedulers, interfaces, and tests.
• Allows reviewers to reject one impact with rationale, inspect revised plans, and approve only the reviewed snapshot.
• Applies approved packages in an isolated workspace, runs verification, and produces a hash-bound proof certificate containing citations, diffs, decisions, results, residuals, and exceptions.
• Uses GPT-5.6 for schema-constrained obligation extraction and redline drafting when an API key is configured.
• Includes deterministic offline fallback using synthetic data and no credentials.

Inference ChangeProof appears to be a compliance or change management tool that integrates AI for automation while maintaining auditability through traceability and versioning.

The author states:

• ChangeProof began with a fictional lender responding to an RBI complaint-handling requirement changing from 30 days to 21 days.
• It was built as an evidence-backed change workflow, not a report or spreadsheet.
• The tool aims to turn external obligations into human-governed, verifiable organizational changes without pretending AI replaces legal judgment or accountable approval.

Inference Positioning has evolved from a hackathon prototype focused on compliance automation to a system emphasizing traceability, governance, and auditability of change processes. It is positioned as a tool for regulated industries or organizations requiring robust compliance workflows.

The description states:

• The initial use case was a fictional lender responding to an RBI requirement.
• It targets organizations needing to respond to external regulatory or compliance changes.
• The system supports “organizational change” and “business change packages.”

Inference The target customer is likely a regulated industry (e.g., financial services, healthcare) or enterprise with compliance obligations. The ICP appears to be internal compliance teams, legal departments, or change management units within large organizations.

Not evidenced.

The description states:

• Built with: aws-app-runner, codex, gpt-5.6, openai-responses-api, python, sqlite.
• Uses Python standard-library runtime components, SQLite, SHA-256 content bindings, strict JSON schemas, computed workspace scanner, and responsive reviewer UI.
• GPT-5.6 is integrated via the OpenAI Responses API.
• Repository has 61 passing tests.
• Optional private AWS deployment overlay.
• Full offline workflow runs without private infrastructure or package installation.

Inference The tool is built with a focus on portability, deterministic behavior, and auditability. It uses AI for automation but includes fallbacks to ensure reproducibility and traceability.

Not evidenced.

Not evidenced.

• The system is described as a hackathon project with no evidence of real-world deployment or adoption.
• It relies heavily on GPT-5.6, which may be costly and difficult to scale without clear pricing or infrastructure details.
• The description does not indicate whether the tool has been tested in production environments or with actual compliance teams.
• There is no mention of enterprise features, integrations, or scalability beyond a demo.

Inference The project is in early development. It lacks commercial traction and may face challenges in scaling or integrating into existing enterprise systems.

1. What external requirements have been tested with this system? Has it been used in any real-world compliance scenarios?
2. How does the tool handle edge cases where AI-generated changes conflict with human judgment or domain-specific knowledge?
3. Are there plans to support other AI models or APIs beyond GPT-5.6?
4. What is the roadmap for enterprise features, integrations, and scalability?
5. Has the team considered cost implications of running this at scale in regulated environments?