Carver: Legal data infrastructure for AI agents

TL;DR: ELI (Executable Legal Infrastructure) turns regulations into structured, versioned, provenance-backed obligations that software - not just lawyers - can consume. We're launching the first two capabilities: Obligation Discovery, which maps the law that applies to a given agent, and Liability Assessment, which prices the exposure. Try it at eli.carveragents.ai - web interface or API, either way the agent spec takes almost nothing to create.

‍

The problem

Agents know a lot about the situation they're in. The customer, the transaction, the conversation history, what tools they can call, what they're about to do next. What they don't know is whether any of it is legal.

Which law applies to this action. What obligation it creates. What fact triggered it. What changed since the last time anyone checked.

That information exists. It just lives in regulations, in outside counsel's head, and in a compliance team's backlog - never in a form the agent, or the system around it, can actually query. So teams either slow down and route everything through legal review, or they ship and find out later.

We built Carver because we kept hitting this ourselves, watching agents go into production with rules that had no idea what jurisdiction they were operating in.

‍

What Obligation Discovery looks like

You create an agent spec - through the web interface or the API. It needs very little: what the agent does, what it touches, where it operates. A PRD or a paragraph both work.
Carver determines what the system actually does - the data it touches, the decisions it makes, the messages it sends, the actions it triggers.
It maps that against a regulatory corpus we maintain continuously - acts, the obligations extracted from them, and how both have changed over time.
You get back a traceable chain: agent → action → facts → applicable law → obligation. Not a list of laws that might apply. The specific ones that do, and why.
Liability Assessment attaches statutory exposure to each obligation - a range, tied to the sources producing it, not a made-up risk score.

‍

What's in the first release

Obligation Discovery - PRD or spec in, applicable obligations out, per agent
Liability Assessment - statutory exposure attached to each obligation, rolled up per agent
A continuously maintained regulatory corpus - acts, obligations, and change data, versioned so you can tell what applied then from what applies now
A web interface and an API - same spec, same output, whichever fits how you work
Low-effort agent specs - a name, what it does, what it touches, where it runs is enough to start

‍

Why we think this matters

Compliance tooling, governance frameworks, and security certifications all do real work - they just weren't built for something that ships weekly, serves five jurisdictions from one web form, and is now operating in areas like hiring and lending where the rules took a hundred years to settle.

‍

We think legal context has to become infrastructure the same way identity, permissions, and observability did - something the agent stack consumes directly, not something a human reconciles after the fact.

This is early. We're starting with discovery and assessment because everything else - control mapping, continuous evals, drift detection - depends on getting the obligation right first.

‍

Our ask

If you're running agents in a regulated workflow - payments, healthcare, lending, hiring, anything with real obligations attached - create a spec for one of them, through the web interface or the API, and see what comes back. We want to know where the mapping is right, where it's wrong, and what you wish it told you.

‍

Try it at eli.carveragents.ai →

‍

United States
447 Broadway,
2nd Floor Suite #563,
New York 10013

Listen to Carver Conversations on Moltbook

AI Regulatory OS
Regulatory Intelligence
Regulatory Sources
Regulatory Platforms
Horizon Scanning
Regulatory Monitoring
Intelligence vs Compliance

Podcasts
Knowledge Base
Resources
Glossary
Use cases
Developers
Blog

Home
Legal Infrastructure
Regulatory Risk Intelligence
Technology
About Us

hello@carveragents.ai
General enquiries
support@carveragents.ai
Product support
legal@carveragents.ai
Privacy and legal requests
github.com/carveragents