Custom Claude Skills, working notes, and deep reporting on how artificial intelligence is actually getting work out the door — for legal teams and the businesses they serve. Written and tested by an in-house attorney, in public.
Illustrative figures based on a single sample document per Skill. Real-world savings depend on document complexity, your workflow, and the discipline you bring to verification. Not a representation about results in any specific matter.
Three ways in, all free. Pull a Skill off the shelf and run it this afternoon. Read a Resource to understand why it works and where it breaks. Book a Service engagement if you'd rather have someone do it with you. Everything here is tagged Legal or Business so you can filter to what fits your role.
Prompt packs, chained workflows, and templates — designed to plug into any agent or chat surface. Free to download, MIT-licensed.
Long-form articles, implementation guides, case studies, and a weekly dispatch for busy in-house counsel and operators who actually need to ship.
Hands-on engagements: audit your current stack, design the workflow, train the team, and handle the ethics-compliance paperwork.
When both legal departments lean on frontier models for first-pass review, the redlines converge. The room where humans add commercial judgment gets smaller — and that's where the work has to happen.
What Judge Castel actually held, what changed in practice, and how I built a citation verifier on this site so the same mistake doesn't happen to anyone running the same workflow I do.
I'm building a voice-driven paralegal agent that hooks into the matter queue. When it ships, I'll write about what worked and what didn't.
Work-product doctrine assumes the attorney's thinking lives somewhere private. The tools we use now put a lot of that thinking on someone else's server. Here's how I think about it as an in-house counsel — and what I do about it.
The Bluebook (22nd ed.) was designed for human researchers using Westlaw and Lexis. The next edition has to account for citations that come out of a model's mouth — and the tools that verify them.
A plain-English primer on what 'reasoning' means inside a frontier model — and where I use it on this site versus where I don't.
A complete four-step workflow — transcript ingestion prompt, witness timeline generator, impeachment-material finder, and a single-page cross-exam outline generator.
Witness: R. Hsu, CFO Topics: Q2 revenue restatement, audit committee communications, board deck versioning. Run: full discovery corpus.
I. BACKGROUND — 12 Q II. TIMELINE — 34 Q ▲ Apr 11 email contradicts Mar 28 board deck III. IMPEACHMENT — 9 Q → Ex. 14 (email chain) → Ex. 22 (slide v3 → v4) 41 exhibits linked
A five-stage Bluebook 22e form-check. Extracts every citation from a draft, maps each to its governing rule, flags non-conforming entries, and produces a .docx verification report plus a printable drafting sign-off for the matter file.
Style: Bluepages (BP)
47 citations extracted
Draft: draft_motion_v3.docx
Authority: Bluebook 22e ✓ 44 CONFORMING ▲ 2 NEEDS CORRECTION §II.B #18 — abbrev. BB R. 10.2.2; T6 §IV.A #34 — F.3d→F.4th BB R. 10.3.1; T1.1 ✗ 1 NON-CONFORMING §III.C #31 — missing components BB R. 10.1 action: re-source / rewrite / remove
A plain-language engagement-letter rider disclosing a firm's use of generative AI. US, UK, or EU regional versions anchored to ABA Formal Opinion 512, SRA Principles, or the EU AI Act, plus a 150-word client summary and a countersignature page.
In representing you, [Firm] may use generative AI tools to assist with research, drafting, and document review. A qualified attorney reviews and is responsible for all work product before it leaves this office. → No client data is used to train third-party models. → Prompts and outputs are retained per §5 below. → You may opt out at any time in writing.
Nine chained prompts that walk a frontier model through a first-pass review of an NDA, MSA, or SaaS agreement. Includes confidence gates and an escalation prompt.
§7.2 LIMITATION OF LIABILITY. In no event shall Vendor's aggregate liability exceed the fees paid by Customer in the twelve (12) months preceding the claim.
▲ §7.2 one-sided cap → request mutual cap → carve-outs: IP indemnity, confidentiality breach, gross negligence suggested: "…except for obligations under §11 (Indemnification) and §13 (Confidentiality)…"
A seven-stage workflow from complaint to filing-ready Rule 12(b) motion. Every citation flagged [VERIFY], every fact paragraph-cited. Closes with a 15-item partner-review checklist.
HCP v. Meridian Diversity · 3 counts: fiduciary duty · negligence · unjust enrichment 22 pp · 51 ¶¶ · 5 exhibits
I. ARGUMENT B. Fiduciary duty ▲ element 1 conclusory C. Negligence ▲ duty / econ-loss rule D. Unjust enrichment ▲ contract governs E. SOL — dates & math war-game: 3 pre-empted 1 unanswered → partner every cite [VERIFY] checklist: 15/15 pass
The deterministic first stage of a clause-by-clause contract-review pipeline. Segments a contract into labeled clauses using a configurable taxonomy, then hands the structured output to downstream reviewer agents.
§§ 7 (warranties) §§ 8 (indem + LoL) §§ 11 (data protection) §§ 14 (governing law) Format: .docx · 4 sections
8 entries emitted → c_001 warranty → c_002 LoL (paired) → c_003 warranty → c_004 indemnification → c_005 LoL (paired) → c_006 LoL → c_007 data-protection → c_008 dispute-res. multi-label: §§ 7.1, 8.1 no paraphrase · verbatim
A jurisdiction-neutral drafting assistant for the six-document estate planning package — revocable trust, pour-over will, DPOA, advance directive, living will, and guardian declaration. Loads state-specific substance from an adopter-supplied checklist.
Spouses, 2 minors $3.8M net worth State X (UPC-based) 6-doc package × 2
Phase 1 intake → ethics gates × 5 → 9 sections captured Phase 2 drafting → 12 docs (mirror pair) ▲ 5 [TO BE CONFIRMED] Phase 3 compliance → jurisdictional ✓ → exec memos × 12 → open items listed
Need more than a prompt pack? I take on hands-on engagements with anyone — in-house teams, solo practitioners, and firms of any size — who wants someone who has lived in both the legal side and the tooling side to help them actually ship.
Every engagement starts with a free 30-minute discovery call. If we're a fit, you'll get a fixed-scope proposal; if we're not, I'll tell you and point you somewhere better.
Book a free discovery callI interview your attorneys and ops team, map your matter lifecycle, and deliver a written report with specific tool recommendations, a workflow map, and a forward-looking roadmap.
Scope and price quoted after the discovery call.Pick one practice area — contracts, e-discovery, litigation support — and I'll design and ship the prompts, schemas, and internal documentation your team needs to run it with AI. Includes hand-off training.
Scope and price quoted after the discovery call.Remote workshops for your team. Hands-on sessions, practice-area-specific exercises, an internal "competence memo" template for each participant, and a follow-up window after the session.
Scope and price quoted after the discovery call.Independent review of your AI policy, engagement-letter language, model-use disclosures, and privilege protocols. Delivered as a redlined memo suitable for your ethics committee or operating leadership.
Scope and price quoted after the discovery call.A monthly retainer for teams that want a running conversation — strategy calls, new-tool evaluations, vendor reviews, escalations — instead of one-off projects.
Scope and price quoted after the discovery call.One long piece, three short case notes, and a single recommendation — delivered Wednesday morning, before opening bell.