I built a browser-only AI workflow safety prototype — looking for feedback

I’m building a small public-safe prototype called WorkflowGuard AI.

The idea is simple:

Before a team connects AI to real customer data, internal tools, or production workflows, they should be able to preview whether an AI workflow idea is safe enough to test.

The demo turns one AI workflow idea into a first-step review:

• risk level
• safety snapshot
• safe first test
• human review gate
• validation checklist
• trace note

Live demo:
https://chuangmenghuan-wq.github.io/traceable-ai-workflow-prototype/

GitHub:
https://github.com/chuangmenghuan-wq/traceable-ai-workflow-prototype

Important boundary:

This is not a finished SaaS product.
It is not production automation.
It does not connect to real customer data.
It has no login, no backend, no storage, and no AI API calls.

The current version is intentionally browser-only and static. I’m testing whether the workflow framing is understandable before adding anything more complex.

I’m looking for feedback from developers, builders, and people who have worked with AI tools inside real workflows:

1. Can you understand what this is within 3 minutes?
2. Does the demo feel useful, or too abstract?
3. Would this help a small team decide where to start with AI?
4. Is the public-safe boundary clear?
5. What would make the demo more useful?

I’m especially interested in feedback on the framing, not just the UI.

Thanks for taking a look.