How Law Departments Use Legal AI to Build AI-First Teams
By Caitlin Price, Social Media & Content Marketer•Updated •Published
At its global offsite in Montreal, Okta's legal team ran a Shark Tank for AI ideas. Mixed groups from across the department pitched improvements, and the top three, chosen by team vote, presented after one night of prep.
Kim Woodward, then VP of Legal and Chief of Staff at Okta, described it on CZ and Friends, GC AI's podcast, where CEO Cecilia Ziniti interviews legal and business leaders about how they run their teams:
It was like Shark Tank basically. We called the panel the council.
She spent the days before running around Montreal assembling gift bags of Canadian candy and maple syrup for the finalists, and the team committed to building the winning idea. Most of the pitches were designed to help the legal team broadly, because the groups had been deliberately mixed across practice areas.
AI for law departments becomes a shared practice when colleagues can reuse company guidance, prepare work consistently, and bring decisions to the right person.
GC AI is a turnkey legal AI platform built for in-house teams and used by over 2,200 legal teams and companies, including 300+ public companies. It supports contract review, research, drafting, and shared company knowledge. Turning those capabilities into a team routine takes ownership, regular training, and a way to keep the underlying guidance current.
Watch Cecilia Ziniti research a privacy question, compare policies, and review NDA redlines in Microsoft Word:
How Law Departments Are Using AI in Practice
A corporate legal team's work reaches beyond contracts. The same department may support a product launch, investigate a regulatory question, prepare board materials, and explain a policy to sales. Start by identifying the work product each colleague needs:
- Contract review and redlining: Compare NDAs, DPAs, and MSAs with the department's preferred and fallback positions. GC AI's Playbooks help prepare a first pass against those standards. Counsel can concentrate review on departures, missing terms, and exceptions that need a decision.
- Legal research: Frame the jurisdiction and factual question, then use GC AI Research to prepare an answer with citations to primary law. The lawyer checks the authorities and their application before using the analysis in advice or drafting.
- Intake preparation: Use AI to summarize a request, identify missing information, and prepare questions for the requester. A clear summary helps the assigned person assess urgency and ownership. Routing and assignment depend on the department's process and configured systems.
- Compliance and regulatory monitoring: Set up GC AI Automations in the web app for recurring monitoring and proposed policy updates. Give the monitoring task a defined subject and scope, then assign someone to assess whether a development changes the company's obligations or guidance.
- First-draft generation: Supply the approved template, matter facts, and drafting instructions for a policy, agreement, or board consent. GC AI for Word supports drafting, review, and redlining in the document where counsel works.
- Business guidance: Turn an approved policy or negotiation position into a plain-language explanation, training exercise, or internal guide. Define which questions colleagues can resolve from that guidance and which require legal review.
A useful workflow becomes a department standard when another teammate can use it with the same sources, instructions, and review expectations.
What Does an AI-First Legal Department Need?
Give the first workflow an owner, make practice part of the calendar, and measure the work through its final review. Those habits let the team improve a process together.
Give the Work a Real Owner
Woodward described a dedicated AI lead at Okta and said that the work needed meaningful responsibility and time.
In your department, that owner might be a legal operations leader or a lawyer with protected time. Their job includes choosing the first workflow, helping colleagues use it, and maintaining the shared instructions. The role also needs access to the people who can resolve legal and operational questions.
Separate those responsibilities even when a small team assigns several of them to one person:
| Responsibility | Owner | Work to Complete |
|---|---|---|
| First-pass preparation | Assigned teammate using AI | Assemble the sources, prepare the output, and identify missing information. |
| Factual confirmation | Relevant business owner | Confirm how the product, service, or business process works. |
| Legal assessment and exceptions | Counsel and the authorized decision-maker | Assess the legal position and obtain any exception approval required by company policy. |
| Knowledge maintenance | Designated resource owner | Update shared sources and instructions when policies, products, or approved positions change. |
At Helix, Chief Privacy Officer, Head of Compliance, and Associate General Counsel KT Farley describes this division of work with GC AI:
Junior teammates now run the checklist prompt first and bring me the output as the predicate for my review.
The checklist gives junior teammates a place to start and Farley a first pass to review. Use recurring corrections to improve the checklist.
Put Training on the Calendar
Okta's legal team made two legal AI days per quarter an annual KPI. That commitment gave experimentation a recurring place in the department's schedule.
If the first legal AI day needs an agenda, this one fits in an hour:
- Pick one live matter: a real contract from this week's queue, reviewed together against a playbook.
- Show one saved prompt: a team member walks through a skill they use weekly and what it replaced.
- Name the gap: each lawyer writes down one task they still do fully by hand.
- Assign one build: the AI owner turns the most common gap into a shared skill before the next session.
- Book the next one: the cadence is the point.
Include the review itself in the demonstration. Open the cited provision, show a correction, and explain why an exception required another person's decision. Colleagues learn how to assess the output as well as how to request it.
One GC used GC AI's Company Profile to explain the business to the platform. She also encouraged newer lawyers to ask it to explain proposed changes and share what they learned in team meetings.
GC AI's legal AI classes include live, attorney-led sessions taught by lawyers with in-house and law firm experience. Your team can learn how to give AI useful context, draft and review in Microsoft Word, build Playbooks around company positions, and assess legal AI ethics. Choose an introductory or advanced session based on the team's experience, or a custom class built around its work. Bring one technique back to the next team session and apply it to a matter in the queue.
Make the Results Visible
Set a baseline before the first workflow changes. Measure the time needed to reach a reviewed deliverable, including source collection, corrections, and the handoff to the requester.
A short weekly review can cover four questions:
- How long did comparable matters take before and after the change?
- Which outputs needed material corrections, such as a missing amendment or the wrong company position?
- Could the recipient use the reviewed deliverable, or did it require another round?
- How much time did the team spend maintaining the shared guidance?
Record what the department accomplished with recovered time, such as clearing a review backlog or taking on a project it had deferred. Track changes in outside counsel assignments and invoices separately. Hours recovered and cash savings answer different budget questions.
Include licenses, implementation, training, and ongoing maintenance in the cost of the workflow. Expand when another teammate can use the process and meet the same review standard.
Company Knowledge Needs to Stay Current
As more colleagues use the workflow, they need access to the guidance behind it: approved templates, negotiating positions, policies, prior decisions, and the reasons for exceptions. Keep that material organized so each reviewer can find the current version and understand when it applies.
Articore Group Chief Legal Officer Jimmy Toy described using an AI-accessible archive of court filings, declarations, and deposition transcripts to check consistency across matters. He checks the answers against the original documents:
you never would have been able to locate those and find the exact passages without that sort of help.
He also gave his legal operations manager a weekly assignment: demonstrate a tool built around the team's knowledge that other people could use.
In GC AI, Files lets teams store and organize documents in reusable collections, share them, and reference sources across chats. Build a collection around a defined task, such as vendor contracting, so the reviewer can find the applicable policy, template, and supporting material together.
Use the Skill Library for the instructions that accompany those sources. A reusable skill can specify what to check, how to structure the answer, and which gaps to flag. A contract-review playbook supplies the preferred positions and fallbacks. The source documents explain the company's rules; the instructions explain how to apply them to the task.
Give each shared resource a maintainer, a version or effective date, a scope of use, and an update trigger. For example, a vendor-review resource might identify commercial counsel as its owner, specify the services it covers, and require review when the company changes its security requirements or negotiating positions.
Keep matter-specific concessions distinguishable from general guidance. A liability position approved for one transaction should retain its conditions and approval record. The next reviewer needs to know whether it establishes a reusable position or records an exception.
When guidance changes, replace the old version in the active collection and preserve its history under the company's records practices. Update the associated skill and playbook as needed. The resource owner remains responsible for keeping those materials current as the business changes.
From Business Request to Reviewed Response
A contract review puts that shared guidance to work. Start with the business request, the agreement and amendments, and the company's negotiating positions. Use those sources to prepare the redline and reply, and record the decisions made during review.
GC AI Agent Connectors can bring information from a user's connected email and file applications into the chat and support actions such as drafting and sending a reply. Each person connects their own accounts, and administrators can disable individual connectors for the organization.
For a recurring contract-review workflow, use that access to gather the request, relevant files, and correspondence. Confirm that the source set includes the applicable agreement and amendments. Obtain anything missing before treating the record as complete.
Apply the department's review instructions to identify provisions that depart from its positions. For each issue, request the supporting passage, a proposed change, and any business or legal decision still needed. Check each recommendation against the cited language.
A requested service commitment, for example, may require an operational answer before counsel can approve the wording. The business owner confirms whether the company can perform it. Counsel assesses the contractual effect, and the authorized decision-maker resolves any exception. The teammate preparing the draft records those decisions in the reviewed work.
Prepare the redline and the reply from that same record. Check that the message describes the changes accurately and distinguishes agreed points from open questions.
Connector action settings include Needs approval, Always allow, and Block. Keep external sending on Needs approval for this workflow. Permission to execute a send is separate from the company's approval of the legal answer.
After sending the reviewed response, save the decision with the matter. If it changes general guidance, send the proposed update to the resource owner. That final step gives the next reviewer the benefit of what the team learned.
Where Does AI Fit in the Legal Software Stack?
Map where the completed work belongs. Your contract system may hold executed agreements and approval records, your matter system may track ownership and deadlines, and your finance system may track outside counsel spend. Decide which system remains authoritative for each record.
GC AI can support analysis and drafting alongside those systems. Test the particular connection or handoff you need, including whether the next person receives the document, source references, and approval record. A useful answer in a chat still needs a reliable destination.
Before expanding access, review the platform terms and settings against your organization's requirements for confidentiality, retention, access, and permitted use. Include the connected applications in that review, and give colleagues clear instructions about which information they may use and share.
The team keeps the judgment calls: which risks to accept, when to escalate, when a deal needs a human no.
Start With a Workflow the Team Can Share
Choose one recurring task with an identifiable reviewer and a useful endpoint. Bring a recently reviewed agreement, the applicable company guidance, and the original business request to a GC AI demo. Ask for an issue summary, proposed redline, and reply that your team could review together.
Have the people who confirm the facts, assess the legal position, and maintain the guidance evaluate the result. Their corrections will tell you what the workflow needs before another teammate takes it on.
Frequently Asked Questions
Can a Small Law Department Use AI Without a Data Science Team?
Does an AI-First Legal Department Need to Replace Its CLM?
How Should AI Change the Department's Use of Outside Counsel?
What Should a GC Ask Vendors to Demonstrate?
Related articles
Resources·Updated Oct 8, 2026What Is a Legal Engineer? The In-House Version of the JobA legal engineer turns repeated legal judgment into a workflow others can run. See the in-house role and how to build your first workflow.
Resources·Updated Sep 4, 2026How Can AI Improve Legal Department Efficiency?AI gives legal teams 14 hours back per lawyer per week and cuts outside counsel spend 14%. The field guide for the GC walking into budget conversations.
Podcast·Published Jun 8, 2026Not Admin Anymore: How Marvell's Two-Person Legal Team Cut Contract Time By 73%Marvell Legal Ops went from one cost center and manual everything to 15-minute SOWs and 73% faster contracts. Sruthi Kosuri and Cindy Prabhakar share how they made it happen.
Podcast·Published May 27, 2026Why Judgment is the New Superpower for In-House Lawyers in the Age of AI Former Unity CLO Anirma Gupta on why judgment is the real job of every in-house lawyer, and how AI does and doesn’t affect a GC’s role.
Build practical legal operations skills with legal AI
See how GC AI supports this work, or build the skills with a practical class for in-house legal teams.



