Preuvance: Instant AI Assurance, Evidence by Evidence

The description states that Preuvance supports a three-step workflow: Prompt, Scan, Prove. It is described as an "evidence-by-evidence ledger" with explicit Declared, Detected, Missing, and Proven semantics.

• Prompt: A user describes an AI system and its context.
• Scan: Supported dependency manifests are parsed in the browser; optional local scan provides a redacted digest.
• Prove: Preuvance assembles a living evidence register that separates Declared, Detected, Missing, and Proven items.

The system uses GPT-5.6 for bounded reasoning tasks such as extracting structured facts, classifying context, and identifying gaps. It also incorporates deterministic TypeScript rules to normalize evidence, enforce allowed states, require reviewer metadata for Proven items, and calculate documentary coverage separately from regulatory readiness.

It stores only evidence metadata behind tenant-aware access controls using Supabase, with no file content stored by the workbench. Browser-side scanners recognize formats like package.json and requirements*.txt, sending explicit digests rather than raw manifest content.

Inference The product appears to be a developer-facing tool for AI governance documentation, not an end-user compliance solution or certification engine.

The description states that teams adopting AI are asked difficult questions about providers, models, controls, and evidence. Preuvance aims to turn scattered information into one reviewable dossier without pretending that AI-generated conclusions are legal certifications.

It positions itself as a documentation assistant for AI governance, not a compliance oracle or certification tool. The authors emphasize that the resulting dossier can be revisited and exported for review workflows, but it does not claim automatic legal compliance.

Inference Preuvance is positioned as an evidence-gathering and organizational tool, not a compliance or certification product. It focuses on transparency and traceability rather than automation of governance decisions.

The description does not name specific customer segments or personas. However, it implies that the target audience includes teams adopting AI systems who need to answer questions about providers, models, controls, and evidence.

It suggests a use case for developers or technical teams working on AI projects where documentation of dependencies, control mechanisms, and risk gaps is required.

Inference The ICP likely includes technical teams in organizations implementing AI systems, particularly those needing structured documentation of their AI governance practices.

There is no evidence provided about pricing, monetization strategy, or business model. The description does not mention any commercial offerings, subscriptions, or revenue streams.

Not evidenced

The project uses:

• GPT-5.6 for bounded reasoning tasks
• Browser-side scanners for dependency manifests (package.json, requirements*.txt)
• TypeScript rules to enforce evidence states and reviewer metadata
• Supabase for metadata storage with tenant-aware access controls
• React, Next.js, Tailwind, Cloudflare Workers, PostgreSQL, SQLite, Remotion

It includes:

• Explicit consent for local scan handoffs
• SHA-256 integrity hashing
• Model tracking in the assessment pipeline
• Deterministic schemas and evidence-state invariants
• Event history and optimistic revisions

Inference The technical stack suggests a modern SaaS or developer tool architecture with emphasis on privacy, traceability, and deterministic validation.

The description states that Preuvance existed before the Build Week event but was focused on new features during the hackathon. It includes a demo repository and testing instructions, but there is no evidence of actual users, customers, or adoption beyond the prototype.

It mentions:

• A downloadable sample PDF dossier
• Local running instructions with OpenAI API key requirement
• Demo workflow using fictional values

Not evidenced

The description does not mention competitors or market positioning relative to other AI governance tools. It focuses on its own internal design and functionality without reference to existing solutions.

Not evidenced

• No commercial traction: The project is described as a hackathon prototype with no evidence of revenue, customers, or adoption.
• Self-reported only: All claims are unverified and based on the author's own description.
• Limited scope: The tool is designed for AI governance documentation, not compliance certification or automation.
• No pricing or monetization model: No indication of how it would be sold or monetized.
• Demo-only functionality: Testing instructions point to a local demo environment, not a production-ready system.

Inference The lack of any commercial evidence makes it difficult to assess viability as a product or business.

1. What is the actual use case for this tool beyond the hackathon prototype?
2. Has there been any real-world testing or feedback from potential users?
3. How does the team plan to monetize or scale this solution?
4. Are there any existing customers or pilot programs?
5. What are the technical limitations of the current implementation that would prevent production deployment?