OSGB AI Pro: Occupational Safety Inspection Copilot

The description states that OSGB AI Pro is an occupational safety inspection copilot designed to support one clear workflow:

• Inspection Notes → Findings → Legal References → Corrective Actions → Expert Approval → Management Report

It is described as a decision-support and productivity tool, not a replacement for expert judgment. AI-generated outputs remain subject to professional review and approval.

The system organizes workplace inspection information into:

• Structured safety findings
• Linked legal and regulatory references
• Proposed corrective actions
• Approval and review records
• Document-based evidence
• Management-ready reporting data

It is built using Python, FastAPI, SQLAlchemy, Pydantic, and relational data storage. The backend follows a layered architecture (Router → Service → Repository), with modules for findings, legal sources, approvals, corrective actions, documents, organizations, and workplaces.

The product is described as being designed around a traceable workflow, where each step preserves accountability and human involvement.

The author positions OSGB AI Pro as:

• A productivity tool that helps occupational health and safety professionals work faster.
• A decision-support system, not an automated replacement for expert judgment.
• A platform that maintains legal traceability, human approval, and accountability at the center of its workflow.

It is described as a copilot, implying it supports rather than replaces human expertise. The goal is to help professionals work faster while preserving legal compliance and professional responsibility.

The positioning has evolved from an initial concept developed with ChatGPT and Fable 5 into a more structured, modular backend system during Build Week using Codex and GPT-5.6.

There is no indication of prior commercial use or market traction beyond the synthetic demo data.

The target customer appears to be occupational health and safety (OHS) professionals who conduct workplace inspections and need to generate structured reports from unstructured field data.

The system supports workflows involving:

• Inspection note-taking
• Legal reference linking
• Corrective action planning
• Expert review and approval
• Management reporting

It is designed for use in occupational safety organizations, potentially including those in regulated industries such as manufacturing, construction, or healthcare.

No specific industry vertical or company size is mentioned. The ICP (Ideal Customer Profile) is inferred from the stated workflow and use case but not explicitly defined.

There is no evidence of a business model or pricing structure in the description.

The author states that the product is built as a decision-support and productivity tool, not a commercial offering. The demo uses only synthetic data, and there is no mention of revenue streams, licensing models, subscriptions, or paid features.

No customer acquisition strategy or monetization plan is described.

The system is built using:

• Python
• FastAPI
• SQLAlchemy
• Pydantic
• Git and GitHub for version control
• Automated backend tests and API smoke tests

It follows a modular backend architecture with clear separation of concerns:

• Router → Service → Repository

Modules include:

• Findings
• Legal sources
• Approvals
• Corrective actions
• Documents
• Organizations
• Workplaces

Cross-module operations are routed through public service interfaces, and legal-reference validation is kept within the legal-sources domain.

Key technical improvements during Build Week included:

• Function-scoped database dependency handling to fix intermittent race conditions
• Prevention of direct access to legal-source models from unrelated modules
• Modular backend enhancements and test suite completion

The author notes that all development work was kept in small, reviewable Git commits, distinguishing pre-existing work from Build Week additions.

There is no evidence of real-world traction or adoption. The demo uses only synthetic workplace data, and no customer base, user feedback, or performance metrics are provided.

The project exists as a conceptual prototype, not a production-ready product. It was extended during Build Week but remains in early development stages.

No revenue, ARR, headcount, or funding rounds are mentioned.

There is no evidence of competitors or competitive landscape described in the project write-up.

The author does not reference existing tools or platforms for occupational safety inspection management or AI-assisted compliance workflows. No market analysis or differentiation strategy is provided.

• No real-world usage: All data used is synthetic; no evidence of actual deployment or pilot testing.
• Unproven commercial viability: No business model, pricing, or customer traction described.
• Limited scope: The product focuses on a single traceable workflow and may not scale beyond that.
• AI transparency concerns: While the system emphasizes traceability, there is no indication of how it handles edge cases or AI-generated errors in real-world settings.
• Single-person team: The project was built by one individual (Hakan G.), which raises questions about scalability and long-term maintenance.

1. What specific occupational safety domains or industries are you targeting, and how do you plan to validate your solution with real users?
2. Have you conducted any user interviews or usability testing with OHS professionals?
3. How will the system handle edge cases where AI-generated outputs may be incorrect or ambiguous?
4. Is there a plan for integrating with existing workplace safety platforms or systems?
5. What are the key assumptions about regulatory compliance and legal traceability that your system relies on?
6. How do you intend to scale beyond a single developer and build out a sustainable product?