Know what your AI is built on.
Every model, provider, and agent your vendors run on — mapped, cited, and refreshed automatically.
Our thesis
Vendor questionnaires are self-reports.
They tell you what the vendor says about itself.
We read the supply chain itself —
and publish what we find.
What it does
A reference guide, posture anchor, and procurement profile.
Encyclopedia
Every model, mapped at a glance.
Browse a living encyclopedia of foundation models, inference providers, and the agents built on top — each entry researched, sourced, and refreshed automatically.
Tracing
Trace each request, root to leaf.
Which models the vendor runs. Who their subprocessors are. Where the data lands. A graph you can navigate.
Posture
One number, defensible reasoning.
A score across retention, training use, certifications, and access controls. Every input cited. Audit the verdict; don't just trust it.
Security score
Acme Copilot
- Data handling88%
- Certifications92%
- Access controls78%
- Model provenance62%
How it works
Research, map, decide.
We research.
Mantleline runs background research on every vendor in your stack — citations, certifications, supply chain.
We publish.
Every approved finding lands in the public encyclopedia — citations attached, ready for the next reviewer.
You decide.
Review the evidence, score the risk, sign off — or send it back. The work product travels with the vendor.
0
Models tracked
0+
Supply-chain edges mapped
0
Compliance frameworks scored
0
Enforcement layers per request
Type II
SOC 2 Type II ready
Aspirational placements — real coverage as it lands
Featured in TechCrunch (aspirational)
Featured in The Information (aspirational)
Higher-Ed CIO Roundtable (aspirational)
AI Risk & Governance Forum (aspirational)
Get started
If your AI footprint is bigger than your governance program, we’d like to help.