Give the business one front door for legal requests, then let AI do the clerical part: read each request, ask for the missing documents, classify it by type and risk, route it to the right person and send status updates from your matter log. Lawyers keep every piece of advice and every approval.
#Key takeaways
- Most intake pain is clerical: missing context, wrong person, no status. That part suits AI well.
- Start with one front door and a short form before adding any AI; scattered channels defeat any tool.
- AI should classify, ask for missing information and draft status updates, never give legal advice to the requester.
- Measure how many requests arrive complete, how many are rerouted and how often people chase legal for status.
- A shared mailbox with a form and a tracker is enough for a small team with low volume.
#What does legal intake look like for an in-house team?
Requests arrive by email, teams chat, WhatsApp, a hallway conversation and a forwarded thread with no attachment. "can you look at this contract" turns out to mean a renewal, a new supplier, a data-sharing question and a deadline that passed yesterday. A lawyer spends the first reply asking for the signed version, the counterparty name, the value band and who owns the deal.
Then the business asks where things stand, again. Nobody can say without opening five threads. The work that needs a lawyer waits behind the work of finding out what the request is.
#Why are business teams going around legal?
When the queue is slow and opaque, people find shortcuts: they sign the counterparty's paper, or they paste the contract into a public chatbot. 78% of AI users at work bring their own AI tools rather than company ones (Microsoft and LinkedIn Work Trend Index, may 2024). A clear front door with visible status is also a confidentiality control, because it removes the reason to go around legal.
#How do you set up a legal front door before any AI?
- Pick one channel: a form, a mailbox or a teams app. Everything else is redirected there.
- Define request types: nda, new contract, renewal, dispute, employment, data protection, regulatory, other.
- Write the minimum information per type: counterparty, document, value band, deadline, business owner.
- Agree routing: which types a contract manager handles, which go to a lawyer, which go to outside counsel.
- Keep one matter log with an owner and a status per request, even if it is a spreadsheet.
#What should AI do in legal intake, and what should it not?
AI is useful at the edges of the queue. It can read a free-text request and its attachments, pick the request type, check the minimum information and ask the requester for what is missing in plain Arabic or English. It can suggest a route with the reason, and draft a weekly status update from the matter log for each business owner.
It should not tell the requester whether a clause is acceptable, whether they can sign, or what the law says. Those answers go through a lawyer. It should not change a matter's status by itself either; the owner does. Anthropic's guidance on building with models is to start with the simplest predefined workflow and add autonomy only when needed (Anthropic, december 2024). Intake is a good example.
#When is a form and a tracker enough?
If a small legal team handles a modest number of requests from a few business units, a form, a shared mailbox and a tracker in your ticketing tool will solve most of the problem. Matter management software with built-in intake is the next step. A custom AI layer earns its place when requests arrive in Arabic and English, across several channels, from many entities, and the same few missing details cause most of the back and forth.
#Where does Stennir fit in legal intake?
Stennir's AI consulting audit maps where legal requests come from, how they are routed and where they stall, and ranks intake against other legal workflows such as NDA triage. If intake comes first, AI implementation builds the classifier, the follow-up questions and the status drafts on your own accounts, connected to your mailbox and matter log, with a named owner approving changes.
AI training covers the business side as well as legal: how to raise a request, what to attach, and why contracts do not go into public chatbots.
We are
Stennir builds the clerical layer of legal intake around your routing rules, in Arabic and English, with your lawyers owning every answer.
We aren't
Stennir is not a legal service and does not answer legal questions from your business teams.
#What should a head of legal do next?
Count last quarter's requests by channel and by how many replies it took before work started. That number shows whether intake or review is your real bottleneck. See how compliance and legal teams use AI on AI for compliance, take the free readiness assessment or book a 30-minute call.
#Common questions
Can AI answer simple legal questions from the business?
We do not recommend it. Even simple questions depend on facts the requester may leave out, and a wrong answer looks authoritative. AI can point people to an approved policy or template your lawyers wrote, and route everything else to a person. Advice stays with your lawyers.
What stays with us if Stennir builds legal intake?
Everything: the accounts, the matter log, the routing rules, the data and the off switch. The workflow runs in your environment, hosted where your rules require. Your team approves changes to routing and templates, and every classification and message is logged so you can see why a request went where it did.
Does legal intake with AI work in Arabic?
Yes. Requests in Arabic, English or a mix are classified the same way, and follow-up questions go back in the requester's language. Stennir tests classification on a sample of your real requests in both languages before rollout, because Gulf business Arabic and formal Arabic read differently.
What could go wrong with AI legal intake?
A request can be misclassified, an urgent deadline can be missed in a long thread, or a follow-up message can sound like advice. The design routes anything uncertain to a person, pulls deadlines out explicitly for the owner to confirm, and uses message templates your lawyers approved.
