AI is now woven into the daily work of an in-house legal team: the first-pass contract review, the fast answer the business is waiting on, the regulatory question that lands without warning. Knowing how to direct it has become core legal competence, the same tier as reading a term sheet or spotting the landmine in an indemnity clause. And it rewards practice: the lawyers who work at it pull ahead, and as the models keep improving, that lead compounds.
Cecilia Ziniti, the three-time general counsel who built GC AI and has taught this to thousands of in-house lawyers, is precise about where the line sits: AI never takes over the judgment layer, the read-and-decide work that is the job.
What it changes is how much you can do with that judgment, and how fast. The skill of directing it comes down to four competencies, each one a move you already make, now aimed at a machine. And you build them the way you built every other lawyering skill: reps, on real work. It is why GC AI’s 101 class drops you into a live contract in the first session.
She put the question to Gagan Biyani on CZ and Friends, GC AI's podcast where we interview legal leaders, technologists, and operators shaping how modern companies work and scale.
Biyani has spent 15 years building places where adults learn: he runs the live-learning platform Maven and co-founded Udemy, and Cecilia teaches GC AI’s classes on Maven, so the two compared notes as builders in the same industry. When she asked what AI skills for lawyers look like in a world where the models improve monthly, he reached for a bicycle.
You can read a book about bicycling. You can watch a YouTube video about it. You can read biking regulations. You can debate the safety of bikes. You can lobby helmet manufacturers. But the reality is if you don’t get your butt on the bike, you’re not going to learn how to ride a bike.
AI skills for lawyers are riding skills. An in-house lawyer builds them on live contracts, real board questions, and this week’s regulatory scramble, and they compound the way any legal skill compounds: through reps with feedback.
The full conversation covers adult learning, attention spans, and why AI adoption is change management in disguise. The riding lessons distill into four competencies, and in-house lawyers build them first.
GC AI is the enterprise legal AI platform built for in-house teams, used by 2,000+ legal teams across 53 countries (Aug 2026), including the legal departments at Liquid Death, Arc’teryx, and Tipalti. Each skill below maps to a product surface you can practice on directly: Easy Prompt turns plain language into structured legal prompts, the Skill Library stores what works as reusable workflows, and GC AI Classes teach the full stack with California CLE credit.
What AI Skills Do Lawyers Need?
AI skills for lawyers are the four competencies that turn a general-purpose technology into reliable legal work product: framing a task, supplying context, evaluating output, and converting what works into reusable workflows.
Regulators already treat the baseline as part of the job. Comment 8 to ABA Model Rule 1.1 folds technology into the duty of competence, and the State Bar of California’s generative AI guidance, updated May 2026 to address agentic AI, expects lawyers to develop a reasonable understanding of the AI systems they deploy on client matters.
Each of the four builds on something the profession already runs on: precise instructions, careful reading, and healthy skepticism. The instincts transfer; the counterpart is what’s new.
GC AI calls the four the In-House AI Skill Stack:
Frame the task. Delegate to AI the way you would brief the world’s best legal intern.
Feed it context. Company facts, standard positions, and the document itself change the answer.
Grade the output. Apply the same read you give a first-year associate’s draft.
Save what works. Turn a good prompt into a reusable Skill your whole team runs.
The In-House AI Skill Stack: 4 AI Skills for Lawyers
None of the four requires code, and each has a place to practice it inside GC AI.
The free 101 class teaches the whole stack in 75 minutes; Easy Prompt handles the framing, the Skill Library saves what works, and Files keeps your company context in reach. Here is how each one works, and where it lives in the product.
Skill 1: Frame the Task Like a Brief to the World’s Best Legal Intern
Task framing means stating the role, the deliverable, and the standard before the request. GC AI’s 101 class teaches it with a casting direction: channel your inner Meryl Streep, exceedingly clear about expectations, working with an intern who is smart, eager, and understood the assignment. Vague delegation produces vague work product from humans and machines alike.
Here is the shape of a framed request for a vendor MSA review:
You are supporting the general counsel of a 400-person SaaS company. Review this vendor MSA against our standard positions on limitation of liability and indemnification. Return a table of clause, risk, and recommended redline. Quote the exact contract language for each finding. Flag anything you would escalate to the business owner.
Purpose-built platforms shoulder part of this skill for you. Maury Bricks, General Counsel and Secretary at ARKO Corp, described what that feels like in practice on CZ and Friends:
I type in ‘please redline this document’ and it’s like, did you mean you wanted to know these 40 things? And I’m like, yes, that’s exactly what I wanted.
That is Easy Prompt doing the framing at the platform level: a plain-language ask goes in, a structured legal prompt comes out. The skill still matters, because the lawyer who knows what a well-framed task looks like can tell in seconds whether the structure fits the matter.
Skill 2: Feed the Machine the Facts of Your Company
Context turns a memo about the law into a memo about your company. The 101 class compresses this into three words: context is king. A useful mental model is Company-Context-Task. State who you are, what constraints apply, and what you need, in that order. The same indemnification question earns different answers for a medical-device company and an ad network, and the model only knows which one you are if you say so.
In GC AI, context works as infrastructure. Attach it once and it rides along. Files holds your reference library as permanent collections, analyzing up to 1,500 pages at once, and a Custom Company Profile carries your team’s voice, templates, and standard positions into each chat. Attach the vendor’s DPA and your data-protection standard together and the analysis starts from your positions, on your paper.
The cheapest prompt is the one you never have to re-ask. Context is how you stop re-asking.
Skill 3: Grade the Output Like Opposing Counsel Drafted It
Evaluation is professional skepticism applied at AI speed: confirm the citations exist, confirm the quoted language appears in the document, and run the send test, asking whether you would put this work in front of your CEO under your own name. Courts keep a running record of what happens when lawyers skip this step; GC AI tracks the sanctions in AI hallucination legal cases.
Platform choice shows up in this skill measurably. GC AI’s In-House Legal Bench, published May 2026, scored four AI platforms across 100 in-house tasks against 1,200+ attorney-developed criteria:
GC AI: 86.8%
ChatGPT (GPT-5.5): 79.8%
Claude (Opus 4.7): 68.4%
Gemini (3.1 Pro): 57.5%
Grading still belongs to the lawyer, and the platform can make the grading faster. Exact Quote cites at the character level, so verifying a finding takes one click to the quoted language. Trisha Mauer, VP of Legal at Tonal, ran the comparison herself:
I’ve compared against ChatGPT, GC AI gives more comprehensive responses appropriate for a lawyer to use.
Skill 4: Save What Works as a Reusable Skill
The fourth competency converts individual wins into team infrastructure. A tested prompt becomes a saved workflow, the workflow becomes a shared standard, and the standard runs on a schedule. This is the skill that compounds, and it is the one the generic advice stops short of showing.
Biyani named its career version “the underlying skill of staying fresh,” the habit that protects a lawyer through whatever the technology does next.
In GC AI, this competency has a literal product surface. A Skill is a set of reusable instructions, built in plain language, that produces consistent results on a task you repeat: objective, inputs, intake questions, workflow steps, output format, and the always/never rules that encode your judgment. The full build process, from intake questions to phase gates to testing, is in how to build Skills in GC AI.
Share a Skill read-only and the whole legal team runs your best workflow identically; schedule it with Automations and the quarterly vendor terms check runs itself. You don’t have to start from a blank prompt, either: the new Regulatory Monitoring Skill Creator builds a standing compliance-tracking agent scoped to your jurisdictions and risk areas in a few prompts.
The lawyers a year from now with a library of twenty tested Skills started this quarter with one.
How Do Lawyers Learn AI Skills?
Reps with feedback beat reading, and structured classes compress the reps.
GC AI’s classes have trained 8,000+ in-house lawyers using the IY Loop: the instructor demos the technique, the group practices together, then you run it on your own matter.
The 101 class covers prompting fundamentals in 75 minutes and carries 1 hour of California CLE credit; the 201 class goes deep on how the models work and advanced use cases. Both are free, taught by former general counsels, and lawyers who complete them adopt AI 3 to 5 times faster.
See the class format and schedule at GC AI Classes.
For the wider training options, including law-school programs and certifications, see AI courses for legal professionals.
To start practicing today, AI prompts for lawyers collects 15 templates in-house teams run daily.
Gagan made the employer math blunt:
If you had someone spend six months learning AI and you knew they were only going to stay for one year afterwards, you would still get enough productivity over that year that follows that six months.
Six months is his conservative case. A lawyer who takes the 101 class this week bills the payback period in days.
One Tuesday Morning, Four Skills
Picture the stack running together. At 8:40 a.m. a DPA arrives from a new analytics vendor, flagged urgent by the head of data.
The associate GC opens GC AI, an enterprise-grade legal AI platform, and frames the ask in Easy Prompt: role, deliverable, escalation rule. She attaches the vendor’s DPA plus the company’s data-protection standard from Files, so the review runs against her positions.
The analysis comes back with Exact Quote citations; she clicks three of them, confirms the language, and catches one sub-processor clause worth a business conversation.
Before the 9:30 stand-up, she saves the whole flow as a Skill and shares it read-only with her team. The next vendor DPA takes eleven minutes, and it takes eleven minutes for whoever picks it up.
That morning is the answer to the question the bicycle quote raises. You get on the bike by riding a real matter, and the legal AI platform remembers how you ride.
Start With One Task You Repeat Weekly
Pick the task, frame it, feed it your documents, grade the output against your own standard, and save what works. That is the whole on-ramp.
Frequently Asked Questions
Do Lawyers Need to Know How to Code to Build AI Skills?
No. AI skills for lawyers are delegation and judgment skills: framing a task, supplying context, evaluating output, and saving what works as a reusable workflow. GC AI’s Skills are built in plain language through a conversational interface, and the 101 class teaches the full stack with California CLE credit.
How Long Does It Take a Lawyer to Learn AI Skills?
A working first pass takes an afternoon: GC AI’s 101 class runs 75 minutes and carries California CLE credit. Fluency compounds with daily use, and lawyers who complete GC AI’s classes adopt AI 3 to 5 times faster, a pattern drawn from the 6,000+ in-house lawyers the classes have trained.
Will Lawyers Who Build AI Skills Replace Lawyers Who Skip Them?
The advantage runs toward the lawyers who build the skills. Maven CEO Gagan Biyani told lawyers on CZ and Friends to learn the new tools and stay at the cutting edge, because those are the lawyers who stay relevant. AI absorbs first drafts and first-pass review while judgment, negotiation, and business context stay with the lawyer; GC AI’s analysis of whether AI will replace lawyers covers the full picture.
What Is the Best Way for In-House Counsel to Start Building AI Skills?
Pick one task you repeat weekly, run it through a framed prompt with real context, and grade the output against your own standard. Pair the reps with instruction: GC AI’s free, CLE-eligible classes teach the full skill stack, and the 14-day free trial gives you a place to practice on your own documents.
