The enquiry nobody answered
A new enquiry arrives at 21:40. It is answered on Thursday, by which time the caller has instructed someone else.
See how the work is measuredQuesto esempio è in aggiornamento. Il testo qui sotto è ancora in inglese.
Example · LegalIntake and matter updates agent
Every enquiry answered, no advice given
An agent that replies to a new enquiry, collects the facts a lawyer actually needs, books the call inside your rules, and records the update on the matter file.
Illustrative scenario — not a real client result
New enquiries · 08:10
- Answered
3 enquiries answered
08:11 · Conflict check run before any name is repeated
- Answered
Consultation booked
08:14 · Fixed-fee first call, 1 hour, Thursday
- Person
“What are my chances?”
08:16 · Legal advice · with the solicitor
Legal advice stops at the agent
Intake and matter updates agent · Legal
- Sourced facts
- A lawyer decides
- Conflicts checked first
01 — A typical day
Three small failures that add up.
What legal owners tell us goes wrong in an ordinary week.
Deadlines living in one person's head
A date is known, entered once, and not checked again. The firm finds out when the other side files first.
See the malpractice pictureA question that is already advice
“Should I sign?” is a legal question. A general assistant answers it cheerfully, and the firm owns the answer.
See the handover rule01 — A typical day · the evidence
What the numbers actually say.
Each figure opens its source, year, sample and limits. Click a badge to check it.
01
Source
ABA Legal Technology Survey 2024, as reported by LawNext
2025
Method and sample are recorded in the source registry.
Read the source for its scope and limits.
Open the source30%
of private-practice lawyers say their office uses AI tools
ABA Legal Technology Survey, 2024 edition, 512 attorneys. An adoption figure, not a measure of results.
ABA Legal Technology Survey 2024 · 202502
Source
ABA Legal Technology Survey 2024, as reported by LawNext
2025
Method and sample are recorded in the source registry.
Read the source for its scope and limits.
Open the source18%
of solo practitioners report using AI
The same survey. Solo and small firms lag midsize ones (30% at 10–49 attorneys).
ABA Legal Technology Survey 2024 · 202503
Source
Texas Lawyers' Insurance Exchange, scheduling errors and malpractice claims
2015
Method and sample are recorded in the source registry.
Read the source for its scope and limits.
Open the source28.49%
of malpractice claims traced to scheduling errors
ABA claims profile for 1996–1999, as summarised by a malpractice insurer. Historical, not a current rate.
Texas Lawyers' Insurance Exchange · 201502 — What the helper does
From your records to a decision you own.
The agent answers a new enquiry in the firm's own words, asks the questions the intake form asks, and runs the conflict check before anyone is told the firm can help.
It books the consultation inside the firm's rules, sends the engagement information, and hands the file to a lawyer with a plain summary of what the caller said.
Every legal question stops. The agent says a solicitor will answer, and routes the message to the person who owns the matter.
6 steps · 2 decisions you own
Reads from
- Web form
- Phone notes
- Practice management
Text version of the flow
- 1. Read the enquiry — Who, what, when, where.
- 2. Run the conflict check — Before anyone is helped. Human checkpoint: Conflicts are cleared by a person.
- 3. Answer in the firm's words — What happens next, and what it costs.
- 4. Book the consultation — Inside the diary rules.
- 5. Stop at advice — Opinion, strategy, outcomes. Human checkpoint: A lawyer answers legal questions.
- 6. Update the matter file — One record, one version.
02 — What the helper does · a simulated example
One simulated example, step by step.
Use the buttons or the left and right arrow keys. Nothing here is a real customer.
Simulated example
Harlow & Finch, a two-partner employment firm
Web form and email · 08:10
A firm of two partners and one paralegal. The agent answers enquiries outside office hours. Every message is simulated; no real client data is used.
This is Harlow & Finch's AI assistant. I can take your details and book a first call. I can't give legal advice. Reply HUMAN at any time.
03 — What you decide
The helper prepares. You decide.
The agent runs the front desk. A lawyer runs the matter.
You approve the wording
Every reply comes from text the firm has signed off, with no improvised explanation of the law.
Conflicts stop the queue
A name match pauses the enquiry until a person clears it — the agent cannot overrule a match.
A lawyer takes over
Merits, deadlines, strategy and documents go to the named fee-earner with the full thread.
You can pause it
Stop intake for a matter type, a partner or the whole firm, and the form falls back to a person.
Without your approval · The agent never advises, never assesses merits, never states a deadline, and never resolves a conflict. Those four are the whole boundary.
04 — What changes when a safeguard is missing
Documented cases, not scare stories.
Three documented cases, each with its source. Two concern AI-generated text; one concerns a diary. None is offered as a firm-level result.
Six fictional cases in a filed brief
Lawyers filed a brief citing six nonexistent decisions produced with ChatGPT, then filed purported copies of the opinions when challenged. The court sanctioned them and required notice to the judges named in the fake citations.
What the helper would do differently
The agent drafts intake notes only. No authority, quotation or citation is generated for a filing, and anything a model proposes is checked by a lawyer against the original.
$5,000
sanction on the lawyers and firm
Corrected after filing, sanctioned anyway
A Massachusetts attorney filed memoranda containing fictitious AI-generated citations. The court found no intent to deceive and the errors were corrected, but still sanctioned the failure to make a reasonable inquiry before filing.
What the helper would do differently
Fixing an error after it reaches the court is not the safeguard. The check happens before submission, and the agent's output is never treated as a verified source.
$2,000
sanction despite correcting the record
Two lawyers left; the deadline did not move
Pro bono attorneys left their firm without telling the client or arranging a substitution, and the local lawyer of record did not act on the order denying relief. The appeal period expired; the Supreme Court held the abandonment could excuse the default.
What the helper would do differently
The agent writes every deadline and owner to the matter record, and a departing lawyer's matters are reassigned before their access is removed. Nothing about a date depends on one person remembering it.
42 days
appeal window that ran out unattended
04 — What changes · the evidence
The rest of the evidence.
Every figure resolves to a source with its sample and limits. No figure without one.
30%
of private-practice lawyers say their office uses AI
ABA Legal Technology Survey 2024, 512 attorneys. Survey self-report.
Source
ABA Legal Technology Survey 2024, as reported by LawNext
2025
Method and sample are recorded in the source registry.
Read the source for its scope and limits.
Open the source512
attorneys in the ABA survey
Private-practice respondents. The published account does not state the response rate.
Source
ABA Legal Technology Survey 2024, as reported by LawNext
2025
Method and sample are recorded in the source registry.
Read the source for its scope and limits.
Open the source18%
of solo practitioners use AI, against 30% at 10–49 attorneys
Same survey, subgroup figures. Not an estimate for all small firms.
Source
ABA Legal Technology Survey 2024, as reported by LawNext
2025
Method and sample are recorded in the source registry.
Read the source for its scope and limits.
Open the source
04 — What changes · your week
The same week, two ways.
Drag the handle, or use the left and right arrow keys when it is focused.
Typical week, beforeTypical week, after
Mon
Eleven enquiries in the inbox. The paralegal triages them between client calls.
Tue
Conflict check done from memory; the partner asks for it in writing later.
Wed
A caller asks whether they have a case; the receptionist answers as best she can.
Thu
Two enquiries are four days old. One has already instructed another firm.
Mon 08:11
Three overnight enquiries answered, conflicts run, two consultations booked.
Mon 08:16
The advice question is in the partner's queue with the full thread attached.
Wed
The paralegal starts with the matters that need judgement, not the ones that need a form.
Thu
No enquiry is older than a working day; the record shows who owns each one.
Illustrative example · not a measured result
05 — Your version
Four questions about your own week.
Keep your answers. They are the brief for the first version we would build for you.
How many new enquiries arrive in a week, and which ones are worth a call?
What do you need to know before you can say whether there is a conflict?
Who may book a consultation, and inside what rules?
What may be written to a client on a matter file, and what may not?
Next
Put your own figures into the calculator below. It runs in your browser, and nothing is sent or stored.
Put your week into the calculatorBring the four answers to the free one-hour call. We will use them to sketch the first version, and you keep the written plan either way.
Talk about your business05 — Your version · your numbers
Put your own week into it.
An estimate, not a promise. The model and its assumptions are printed below.
Your inputs
Web form, email and phone notes together.
count
Facts, conflict details, booking — no advice needed.
%
Reading, replying, checking conflicts, booking.
min
Paralegal or fee-earner time, whichever does intake.
per hour
Evenings, weekends, court days.
%
From enquiry received to a human reply.
hours
Example defaults are typical for a small business. Replace them with yours. Nothing is stored, sent or shared.
Estimated result
IllustrativeFee-earner hours back each month
0–0
- Time released at your rate
- £0–£0
- Enquiries answered outside office hours
- 0–0%
- Time to first reply
- Already immediate
How we calculated this
Routine work handled by the helper: 70% of the share you marked routine.
Time saved per item: 80% of the minutes you entered.
Monthly volume: weekly figures × 4.33; monthly figures used as entered.
Cost: hours saved × your staff cost per hour, in pounds as an illustration.
Reply time: routine items move within the hour while the helper runs. Complex work still waits for a person.
Questions
The things people ask next.
No. Merits, limitation dates, strategy, documents and likely outcomes go to a lawyer. The agent acknowledges the question and routes it with the thread.
No. It runs the check and stops on any match. A person clears the enquiry or declines it; the agent cannot overrule a match.
Yes. The first message says it is the firm's AI assistant, that it cannot give legal advice, and that a person is available on request.
The agent reads dates from the matter record but never calculates or states a limitation period. A lawyer confirms every date in writing.
No. It answers the enquiries that arrive when nobody is free, and passes everything that needs a person to the same inbox the receptionist uses.
Intake data stays in the firm's systems, is encrypted, and is kept on the firm's retention schedule. It is not used to train a model.
Two to three weeks: approved wording, conflict-list import, and a two-week parallel run alongside the existing intake.
Start here
Show us last month's enquiries.
One hour with a week of real enquiries. We will show you which ones the agent would have answered, and which ones must never reach it.