In-house legal teams adopting AI: first use cases
By the ChamberOne team 5 minute read
An in-house lawyer is rarely short of requests. A sales contract by lunchtime, a question about the leave policy, a supplier dispute that nobody wanted, and a board paper due on Friday. Two or three lawyers can end up acting as a help desk for the whole company. AI will not fix that overnight, but a few well-chosen starting points can free some hours for work that needs your judgment.
Key points
- Start by logging where legal's time goes, then choose repeated, low-risk, checkable tasks.
- Contract triage, policy questions and matter intake are sensible first use cases.
- Have a one or two page governance statement ready for the business.
- Measure response times, escalations and corrections, and share the results.
Begin with the queue, not the technology
Before you look at any tool, look at where your time goes. For two weeks, jot down each request: who asked, what kind it was, how long it took and whether it was routine. Most legal teams discover that a large part of the volume is a small number of repeating types.
The best first use cases are repeated, low-risk, well understood and easy to check. Avoid starting with the most complex and sensitive work, however tempting. Early wins build confidence; early failures build resistance.
- Log each request for two weeks
- Start with work that is repeated and low-risk
- Pick work that is well understood and easy to check
- Look at tools before you look at where your time goes
- Start with the most complex and sensitive work
Use case one: contract triage
Incoming contracts are the usual place to begin. An AI tool can read a contract, identify its type and the parties, list key terms such as term, renewal, liability and termination, and flag clauses that depart from your standard positions. The lawyer then spends time on the flags, not on reading every page.
Set this up with a playbook: a short statement of what you accept, what you negotiate and what you refuse. Group contracts into low, medium and high attention. Keep humans in charge of the decision, and sample the low-attention lane to catch the errors that would otherwise go unseen.
Use case two: policy questions
Employees ask the same questions again and again. Can I accept this gift? What is the process for approving a supplier? Who signs this? A question-and-answer tool built on your own approved policies can handle many of these, with links back to the policy text.
The rule is that it answers only from your documents, and it says so when it does not know. Review the answers from time to time, and add a clear way to reach a lawyer. For anything touching discipline, investigations or legal risk, the tool should route to a person rather than reply.
Use case three: matter intake
Requests to legal often arrive as vague emails. A simple intake form, with a few required fields such as the business owner, the deadline, the amount at stake and the documents attached, saves back-and-forth. AI can read free-text requests and suggest a category and a priority, or draft a short summary for the lawyer who picks it up.
Use the form to make the queue visible. Once requests are logged, you can show the business how much work legal handles and where the pressure is, which helps with resourcing discussions.
Governance the business will accept
The business will ask whether using AI is safe. Have a clear answer ready, in one or two pages.
- Approved tools only, assessed on retention, training use, location of processing and sub-processors.
- Rules on what may be entered: public, internal and confidential categories, and what stays out entirely.
- A statement that output is checked by a lawyer before it is relied on, with the checking recorded.
- A named owner of the policy and a review date.
- A log of use cases and any incidents, reviewed by the general counsel.
- Alignment with the company's information security and privacy teams, who should be involved from the start.
Managing expectations of business partners
Some colleagues will expect instant answers. Others will worry that legal is being replaced. Talk to both. Explain what the tool is for: faster handling of routine requests, with lawyers still responsible for advice. Tell them what will not change: complex matters still need a conversation, and some requests will take as long as before.
Set expectations on response times for each category, and publish them. A clear service standard is more persuasive than a promise of speed.
What to measure
- Time from request to first response, by category.
- Share of requests handled through the standard route versus escalated.
- How often a lawyer changed the tool's suggestion, and why.
- Satisfaction among the business users who rely on the service.
- Any near misses where a check caught an error.
Choose two or three measures, track them honestly, and share them with the business. Numbers that show what is and is not working earn more trust than enthusiasm.
Where to be careful
Privilege can be affected by how material is shared and with whom, so involve the general counsel in deciding which tools may see privileged documents. Be cautious with documents that are subject to confidentiality terms with third parties. Do not use tools on material that is subject to regulatory restrictions without checking the position. And remember that regulated industries often have their own rules on technology and outsourcing, so involve compliance early. Check the guidance from your regulator or bar for lawyers' own duties.
A sensible first quarter might look like this. In the first month, log requests and write the one-page governance statement. In the second, pilot contract triage on one contract type, with a lawyer reading every result. In the third, add the policy question tool for a single department and review the answers weekly. At the end, sit down with the business and decide what to keep, what to change and what to stop. A small, reviewed pilot is easier to defend than a company-wide launch.
Budget for the unglamorous parts too: time to write and maintain the playbook, time for sampling and review, and time to talk to colleagues. The tool is the smaller part of the cost.
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
How a small law firm can start with AI without a big project
Pick two low-risk uses, set review rules and measure before expanding, with a list of what to postpone.
5 minute readA 90-day AI pilot plan for a law firm
A week-by-week plan to choose a use case, set measures, run a small trial, review errors and decide.
5 minute readChoosing AI models and firm-level controls: what a practice should decide
Model choice, an off switch and a usage record are three decisions a firm should make deliberately.
4 minute read