Publications

Books from the KAOS team.

Two long-form KAOS books in scaffold for Q3 2026. Two conceptual minibooks already in print at ai4lf.com — the assigned reading every KAOS chapter builds on.

Reading order

The two minibooks at ai4lf.com are the conceptual canon: how a language model actually samples a token, and what makes a system agentic. The two KAOS books then show what to build on top — one for lawyers running an agent, one for developers shipping the system behind it. Read either KAOS book without the minibooks, but expect the assigned-reading callouts to point you back.

LLM Essentials conceptual canon · ai4lf.com Agentic AI design vocabulary · ai4lf.com Vibe Coding for legal professionals · Q3 2026 Building Apps for Python developers · Q3 2026 KAOS curriculum Bommarito · Katz · Bommarito

Vibe Coding with KAOS

An AI Coding Guide for Legal and Finance Professionals

For lawyers, paralegals, analysts, compliance officers, and legal-ops managers who already use Claude Code. By the end of chapter 1 you have classified a folder of PDFs into contracts, memos, court filings, and other, and verified the result against a ground-truth folder. By chapter 6 you have produced a partner-reviewable flagged-issues CSV across twenty contracts on a six-category buy-side rubric (change of control, IP assignment, non-compete, indemnification, limitation of liability, termination) and audited five flags at random against the source clauses. Every chapter ends with commands you run yourself, and chapter 10 grounds every workflow in the ABA Model Rules and the named jurisdictional ethics opinions.

Bommarito · Katz · Bommarito · Q3 2026 — outline available

Building AI Applications with KAOS

A Practitioner's Guide from Library to Production

For Python developers shipping legal-AI features. Move from a one-line parse of an SEC filing to writing your own KaosTool that an M&A associate calls from Claude Code, then publish it as an MCP server other teams can install. The book treats the typed document AST, citation-anchored retrieval, the cost ceiling, and the ABA-aware audit trail as the load-bearing properties — not afterthoughts. Grounded, verifiable, cost-aware applications, not demos.

Bommarito · Katz · Bommarito · Q3 2026 — outline available

LLM Essentials for Law and Finance

How LLMs actually work, for the people deploying them

The conceptual canon. By the end you can read a prompt the way a litigator reads a brief — knowing where the model is most likely to hallucinate, where token limits will silently truncate the record, where a structured-output schema saves you, and where a tool call is the right answer instead. The assigned reading for KAOS Part I (Foundations).

Bommarito · Katz · Bommarito · Print-ready · ISBN-registered · ai4lf.com

Agentic AI in Law and Finance

Recognize, design, govern

The conceptual canon for agents. By the end you can sit in a partner meeting, watch a vendor demo, and tell whether the system is agentic, what its termination and escalation conditions are, who supervises it, and where the audit log lives. The design vocabulary — triggers, intent, perception, action, memory, planning, termination, escalation, delegation — is the language KAOS implements. Assigned reading for KAOS Parts II to IV.

Bommarito · Katz · Bommarito · Print-ready · ISBN-registered · ai4lf.com