AI for legal
AI for legal teams reads your firm's own contracts and tells you which ones end or renew soon and what notice each needs, with the document and clause number beside every date. Time narratives get tagged to matter phases too, so billing questions come back with the query that answered them.
One question, traced, example
4 contracts renew automatically by 20 May 2027, and 1 notice deadline has passed
You asked
Which of our supplier contracts renew automatically in the next six months?
Checked 14 supplier contracts6 steps
-
Read the question0.2s
You want the supplier contracts that roll over on their own between today, 20 November 2026, and 20 May 2027, and when notice would have to be given.
-
Decide whether to query0.1s
The terms were extracted from each contract when it was filed. This is a query on those fields, with no need to reread the documents.
-
Pick the dataset0.3s
Used contract_terms, one row per contract. Every field carries the clause number and the sentence it came from. The firm's own supplier contracts sit under one internal matter.
-
Write the query0.4s
Keep contracts that renew automatically and end inside the window, plus any where no renewal clause was found. Notice deadline is the end date less the notice days.
WITH supplier_terms AS ( SELECT t.document_id ,t.service ,t.auto_renews ,t.term_end_date ,t.term_end_clause ,t.renewal_months ,t.renewal_clause ,t.notice_days ,t.notice_clause ,t.term_end_date - t.notice_days AS notice_deadline FROM legal.contract_terms t WHERE 1=1 AND t.matter_id = 'FIRM-SUPPLIERS' ) SELECT s.document_id ,s.service ,s.term_end_date ,s.term_end_clause ,s.renewal_months ,s.renewal_clause ,s.notice_days ,s.notice_clause ,s.notice_deadline ,s.notice_deadline - DATE '2026-11-20' AS days_to_deadline FROM supplier_terms s WHERE 1=1 AND ( (s.auto_renews = TRUE AND s.term_end_date BETWEEN DATE '2026-11-20' AND DATE '2027-05-20') OR s.auto_renews IS NULL ) ORDER BY s.notice_deadline NULLS LAST -
Run the query0.6s
Query finished against the read-only view. Nothing was written.
Write the answer0.3s
Every date below comes from those 5 rows, each with the clause it was read from.
| Contract | Service | Term ends | Renews for | Notice | Notice deadline | Days left |
|---|---|---|---|---|---|---|
| CTR-07 | Document storage licence | 31 Jan 2027 cl. 14.1 | 12 months cl. 14.2 | 90 days cl. 14.3 | 2 Nov 2026 | passed 18 days ago |
| CTR-03 | Office cleaning | 28 Feb 2027 cl. 2.1 | 12 months cl. 2.4 | 30 days cl. 11.2 | 29 Jan 2027 | 70 |
| CTR-11 | Printer and copier lease | 31 Mar 2027 cl. 4.1 | 12 months cl. 4.2 | 60 days cl. 4.2 | 30 Jan 2027 | 71 |
| CTR-05 | Off-site archive storage | 30 Apr 2027 cl. 6.1 | 24 months cl. 6.2 | 60 days cl. 19.4 | 1 Mar 2027 | 101 |
| CTR-09 | Courier services | 30 Sep 2026 cl. 2.1 | No renewal clause found | n/a | n/a | for a person to read |
Four supplier contracts renew automatically between today and 20 May 2027. CTR-07, the document storage licence, comes first: it ends on 31 January 2027, renews for 12 months, and takes 90 days' notice, so its notice deadline was 2 November 2026, 18 days ago (cl. 14.1 to 14.3). The next two deadlines are 29 January 2027 for office cleaning (CTR-03, cl. 11.2) and 30 January 2027 for the copier lease (CTR-11, cl. 4.2). CTR-09, courier services, has no renewal clause the extraction could find, and its term ended on 30 September 2026, so it is listed for a person to read.
What works now in AI for legal teams
What does not work yet
An extracted date with no clause number beside it is a date someone has to find all over again.
What has to be true of your legal data first
The firm's own contracts are held as searchable text
Extraction cites a document and a clause, so without text it has nothing to cite.
The firm's own contracts are filed apart from client matters
Extraction reads one location, such as the internal matter in the example above. A supplier contract filed inside a client matter moves there first, which keeps client files out of scope.
Time entries keep their narrative, matter type and fee earner
A narrative overwritten with a billing code, or a matter type in free text, leaves nothing to tag against the phase list.
The AI readiness assessment starts with these three checks on your contracts folder and practice management system. It then answers ten of your partners' questions by hand, such as which supplier contracts renew in the next six months.
Regulation and risk for AI in legal work
The sources we read on 20 November 2026, quoted or named.
Legal Services Act 2007
Section 12 covers reserved legal activities.
Checked 20 November 2026.
The ICO's guidance on criminal offence data
The guidance says:
"The UK GDPR gives extra protection to the personal data of offenders or suspected offenders in the context of criminal activity, allegations, investigations, and proceedings."
The page also carries this notice:
"Due to changes made by the Data (Use and Access) Act, this guidance is under review and may be subject to change."
ICO, criminal offence data. Checked 20 November 2026.
Automated decisions are covered once, on the hub page under What every sector shares.
This is a list of the sources we read, with the date. It is not legal advice, and your own adviser decides what applies to you.
When a partner questions an extracted date, they can open the clause it came from and read the wording for themselves.
Where the method is explained
Questions people ask
Can AI read our contracts and pull out the key terms?
Yes, for your firm's own contracts, such as supplier contracts and leases, held as searchable text. Each field is stored with its document ID, clause number and source sentence, so a person can check it before anyone relies on it.
Does the AI give legal advice on what a clause means?
No. It reports what a clause says and where, and a lawyer decides what it means. The notice deadline in the example above is the end date less the notice days, and how the clause itself counts those days still needs a lawyer's reading.
Will the AI see privileged client documents?
No. Extraction reads one internal location holding the firm's own supplier contracts and leases, and client documents and matter files stay out, as the legal analytics page also says. Every answer shows the query it ran.
How does tagging time narratives help price fixed fees?
It tells you how many hours each phase really took. Once each time entry's narrative is tagged to a phase from your list for that matter type, recorded hours add up by phase across closed matters, and the next fixed-fee quote can rest on them.
What happens when the extraction can't find a renewal clause?
The field stays empty and the contract goes on a list for a person to read, as CTR-09 is in the example above. Its term ended on 30 September 2026 and no renewal clause was found. A guessed date would have no clause to cite, and every extracted field has to cite one.
Do your contracts and time entries hold what AI would need?
Ten supplier contracts or leases and one month of time entries let the first call mark which fields have a clause to cite and which narratives fit a phase.