All client work
Case study · Legal / professional services

How a regional personal-injury firm captures after-hours intake without allowing AI to give legal advice.

This regional personal-injury firm operates multiple offices and receives calls around the clock, but reception cannot staff every hour. Kingstone built an AI intake agent to gather the initial facts and recognise when a question requires a lawyer rather than an automated response.

Legal / professional servicesRegional personal-injury firmName withheld by request
Client
Anonymous, name withheld by request
Location
Multi-office, regional
Industry
Legal / professional services
Channels
Inbound voice + web intake

From the client

After-hours calls no longer leave us with just a voicemail. Our attorneys return the call with the incident details already in front of them.
Managing PartnerRegional personal injury firm

The challenge

Before launch, calls that arrived after hours or while reception was unavailable often went to voicemail without the firm collecting the facts needed for a useful attorney callback. That delay created avoidable intake gaps at the moment a prospective client was actively asking for help.

The firm needed to be reachable around the clock without expanding its front desk, and without letting an automated system anywhere near legal advice or case valuation.

The solution

Kingstone deployed an AI intake agent for calls the front desk cannot answer. It captures the incident date, whether a police report was filed, whether the other side's insurer has made contact, and the details needed to book an attorney callback. If a caller asks what a claim is worth, who was at fault, or how long they have to file, generation stops and the caller receives an attorney-approved response written in advance, while the conversation is flagged for a person. It does not estimate case value or give legal advice.

Documented deployment workflow

Initial intake with a defined legal boundary

Callers can share the important facts immediately and reach the right next step, while legal advice and case judgment stay with attorneys.

01Call reaches the firm

The front desk answers when available; uncovered calls enter the AI intake flow.

02Initial facts are collected

The agent asks about the incident date, police report, insurer contact, and callback details.

03A callback is prepared

The captured context is prepared for an attorney callback rather than left in voicemail.

04Legal judgment stops the flow

Case value, fault, filing deadlines, and substantive legal issues are escalated to a person.

The firm also received a dedicated dashboard to review intake activity, urgent flags, attorney handoffs, deployment notes, and the metrics the team tracks.

The results

Figures below are from intake calls handled in the first 90 days of live deployment. Outside-hours share is measured against the firm's published office schedule.

1 in 3
Intake that used to hit voicemail, now captured live

Outside office hours and weekends in the first 90 days, measured against the firm's schedule.

<2 min
Urgent matters flagged to the team

Attorneys see high-priority intake almost immediately instead of discovering it the next morning.

Attorney-ready
Facts gathered before the callback

Incident date, police report, insurer contact, and callback details — with zero legal advice from the AI.

Hard stop on legal judgmentDollar figures, fault, deadlines, and “do I have a case” questions stop the flow in code before the model runs. The AI never values a case or assigns blame.
Voice still leadsMost people still prefer to talk it through by phone; the rest start on the web intake form.
Reachable without expanding the front deskAfter-hours coverage without staffing reception around the clock.

This case study describes a real Kingstone client deployment. The client's name and identifying details have been withheld at its request. Figures shown are from this client's deployment.

See how this workflow could fit your operation.

Book a working session to map intake questions, prohibited topics, and human escalation rules.

Book a Working Session →