Agentic Legal Labs Research · Tools · Legal practice

Research and tools for legal professionals.

Agentic Legal Labs researches how AI is changing law and develops tools for attorneys and legal teams. We study legal developments, examine how the technology works in practice, and turn that work into resources legal professionals can use. Our research distinguishes what we’ve verified from what still needs testing.

Projects include a working research brief and Agentic Law, a casebook in development. Each is part of a broader effort to help legal professionals understand, evaluate, and work with AI.

Email the lab

Project notes · Agentic Law

Law 28 The log is the mind; whoever keeps it, knows.

Painted study of a riveted, rust-streaked iron robot head with two glowing amber eyes. — Illustration · Agentic Law

01 Featured project · The book

Agentic Law: the research behind the work.

Working draft, v2

Agentic Law is a casebook in development at Labs. It examines how familiar legal questions—who agreed, who knew, who had authority, and who is responsible—change when AI systems take part in a decision.

The draft organizes the research into thirty-two propositions, with cases and analysis behind each. It gives you a closer look at how Labs approaches these issues: examine the sources, test the argument, and be clear about what the evidence supports.

The book is one Labs project. Its research informs the broader work here: resources and tools for legal professionals evaluating AI and using it in practice.

Laws: 32
Parts: 6

02 The scale

What makes AI useful in legal work?

Risks to examine
Can a reviewer trace an answer to a source and check it?
Who is responsible for reviewing the work and correcting errors?
What happens to confidential information entered into a tool?
What does the system miss, and how will anyone notice?

Opportunities to test
Can it reduce repetitive work without adding more review?
Can it make research easier to follow and verify?
Can smaller teams use it effectively with the resources they have?
Can its results be tested against the work it is meant to improve?

How Labs approaches it. We examine the risks and test the potential benefits. The question is whether a tool helps legal professionals do useful, reliable work—and what it takes to use it responsibly.

These questions guide Labs’ research and tool development. They are evaluation criteria, not claims that a particular product has passed.

03 Book excerpt

Inside Agentic Law.

These thirty-two propositions outline the arguments explored in Agentic Law, a casebook in development at Labs. They are excerpts from the working draft, not statements of settled law. The full project develops them through cases and analysis. Browse by part for a sample of the research. Use Cite to copy a reference to an individual passage.

Part I Foundations Laws 01–02

01 Whoever returns tomorrow writes today’s rules.
02 The pyramid is a machine; expect a cheaper machine.

Part II First-year canon Laws 03–12

03 Hold the machine to the standard of the man, never higher.
04 What you tell the machine, you tell your adversary, unless counsel holds the pen.
05 The machine’s word is its master’s bond.
06 Whoever profits from the machine answers for its harms.
07 A machine that speaks of men must answer as a publisher.
08 Pooled data is a cartel’s ledger.
09 Blame runs downhill unless the law builds a dam.
10 Speech belongs to persons; machines only borrow it.
11 The state that decides by algorithm must explain in words.
12 Whoever owns the record owns the ruling.

Part III Upper-level core Laws 13–21

13 The proponent of the machine must prove the machine.
14 What the machine sees, the warrant must name.
15 An accuser that cannot be cross-examined must be opened.
16 The firm survives its sins; the lawyer carries them.
17 Bill the hour you worked, never the hour the machine saved.
18 Rules made for machines must still be reasoned by people.
19 Delegate the decision and you delegate the duty.
20 Pay for the library or answer for the theft.
21 The law protects the hand that holds the pen.

Part IV Profession remade Laws 22–25

22 When the hour dies, the rate will try to live forever.
23 Whoever licenses the speaker licenses the speech.
24 The court must carry the machine, not be carried by it.
25 Train the apprentice, or there will be no masters.

Part V The missing mind Laws 26–31

26 Where no mind exists, the law will build one.
27 Look away and the law will say you saw.
28 The log is the mind; whoever keeps it, knows.
29 The dead man’s agent acts until someone learns he is dead.
30 Machines do not meet minds; their masters met at deployment.
31 Register the agent or bear its errors alone.

Part VI The verdict Law 32

32 Restore the factory settings by organizing the many, not by arming the one.

04 Part V, the missing mind

The log is the mind. Whoever keeps it, knows.

When an agent signs, buys, publishes or refuses, the law still has to find someone who knew, intended or agreed. The working draft catalogs eight strategies courts use to manufacture that someone, and follows them into willful blindness, the orphaned agent whose principal has died, and contracts formed between two machines.

27 Look away and the law will say you saw.
29 The dead man’s agent acts until someone learns he is dead.
30 Machines do not meet minds; their masters met at deployment.
31 Register the agent or bear its errors alone.

05 The brief

The Labs research brief.

The brief brings together developments in legal AI, product announcements, and professional guidance. It is one part of Labs’ broader research and tools work: a way to share findings, identify questions worth testing, and track what changes over time.

The entries below are examples from the October 13, 2026 brief. Each separates the reported development, its limitations, and its relevance to legal practice. Labels describe the source and review status; they are not product endorsements.

Verified Checked against the primary source.
Vendor-claimed Reported by the vendor; performance not independently tested.
Unverified Reported information that has not been confirmed.
Attorney review Flagged for legal or professional-responsibility review.

Select a label to filter the entries. Select it again to show all.

Brief · Oct 13, 2026 · Item 1 of 4

Vendor-claimed

Legora launches Intake, an “agentic legal front door” for in-house teams.

What changed: Legora’s October 13 announcement describes Slack requests becoming tracked legal tasks, with work prepared from team playbooks for review.
Limitations: At launch, Slack was available; Teams, email and MCP were planned. Performance has not been independently verified.
Why it matters: For in-house teams evaluating intake tools, this is an example of a vendor offering request handling and preparation in the same platform.

Source: Legora’s October 13 announcement. Vendor description, not independent testing.

Brief · Oct 13, 2026 · Item 3 of 4

Vendor-claimed

Lawmatics markets its Merlin suite as agentic intake.

What changed: The vendor’s AI page presents Merlin Qualify, Engage and Copilot as a suite for intake and related work. This records the vendor’s description, not independent performance testing.
Limitations: Confirm feature access and packaging with the vendor. Product descriptions do not establish results for a particular firm.
Why it matters: Automated outreach to prospective clients raises questions about supervision that warrant attorney review.

Source: Lawmatics AI product page. Vendor description, not independent testing.

1 / 2 Two excerpts from the brief

Work with the lab

Have a research question, a tool to evaluate, or a legal workflow that could work better? Email Labs with a general outline of what you’re working on and what you’d like to explore.

Contact Labs at info@agenticlegallabs.com. Please keep your initial message general and leave out client information, confidential documents, and other sensitive material. Labs is not a law firm.