Law Firm: AI Receptionist case study

For years, the front desk at this small London law firm was managed by one receptionist, supported by two legal assistants. Many of the calls it handled were simple enquiries about directions, parking, opening hours or how to submit documents.

Outside office hours, an external answering service took messages.

Today, AI Receptionist handles routine daytime calls and out-of-hours enquiries, helping the firm reduce its reliance on that service, which cost more than £1,000 a month. And the decision to make the change took just one meeting.

The firm’s been serving clients in London for more than 20 years and is regulated by the Solicitors Regulation Authority. It works across six areas of law: family law, residential property and conveyancing, wills and probate, crime, landlord and tenant, and care proceedings.

Most new business comes through recommendations. The firm has worked with multiple generations of the same families, and its reputation is built on trust and personal service, meaning every call is part of the client experience.

That’s why, when the owners explored using AI at the front desk, they started with the receptionist, whose role was never under threat. The focus was on taking the repetitive calls out of their day.

With six different practice areas, every call needs to be understood before it can be directed. Is the caller a new client or an existing one? Which area of law is involved? Which fee earner should help? Is the matter urgent?

Answering those questions requires experience and judgement. Yet the same person was also fielding the simple enquiries.

Some of the most important calls arrived after the office had closed, often from people who needed help immediately following a stressful event. The previous answering service could take a message, but it couldn’t answer questions or arrange appointments.

Working with BT’s training partner, one member of the firm helped configure AI Receptionist using a requirements form. The system was set up around the firm’s priorities, preferred call-handling process and appointment-booking requirements.

It learned the basics from the firm’s website, including its location, opening hours and areas of practice. Additional information from the firm was then added, and the system was tested before launch.

AI Receptionist is now configured to operate 24/7, identifying the relevant area of law, distinguishing between new and existing clients, and directing callers to the right fee earner. It can collect key details from prospective clients, including the name of the other party, helping the firm carry out conflict checks before a conversation takes place.

It answers common questions, and can book appointments and send confirmations.

Some tasks remain with the receptionist. Decisions about whether a caller should speak directly to a partner require additional questions and professional judgement. An urgent court matter is put straight through to the team.

The phone still rings at the front desk, and the receptionist still answers it.

The difference is which calls get through. Routine requests go to AI Receptionist first, so the calls that reach the receptionist are the ones that need a person.

Out of hours, callers receive a response rather than leaving a message and hoping for a callback.

AI Receptionist also shows how the service is performing. It records every interaction, providing transcripts and reporting that show how many calls were handled, how many were transferred, how many remained unresolved and what was said. A business can use these insights to fine-tune the answers AI Receptionist gives and improve the caller experience.

Small and medium businesses want advice they can trust before they take on AI. BT’s scale, experience and long track record of delivery give them that confidence. Advising customers well is a responsibility we take seriously, so our people go through BT’s own AI training and accreditation.

The help continues once AI Receptionist is up and running. A long-standing BT training partner gives customers videos, materials and support to get the most from it. For a practice that holds privileged client information, one more thing mattered: we don’t use customer data to train AI models.