When the firm's secretary went on sick leave, more than 200 inbound calls a month were left without a dedicated resource. An external call center had been set up, but it never managed to absorb the volume or deliver the quality of reception expected by clients in distress.
Two effects combined, and the second cost more than the first. On one side, the lost calls: the phone ringing out during a hearing, messages never returned, prospects gone elsewhere. On the other, the calls that were answered — but by the lawyers themselves. Picking up between two files, jotting a number down on the corner of a desk, calling back in the evening on the way home from court: several hours a week taken from advocacy, drafting and advisory time. A partner taking a message is not a partner pleading a case.
There was also a deeper reluctance, and it was a legitimate one. The firm's clientele is mostly criminal: people calling after police custody, a summons or an indictment. Calls loaded with anxiety, where the first contact sets the tone for the whole relationship of trust. Handing that reception to an artificial intelligence was far from obvious — on the face of it, it was the perfect counter-example. That objection is precisely what shaped the deployment.
- Calls lost during hearings and police custody, with no structured call-back process behind them.
- Several hours a week spent by the lawyers themselves picking up, screening and calling back, at the expense of billable time.
- Sensitive information not passed on at the right moment: hearing dates, court registry calls, client emergencies.
- No simple appointment-booking system and no prepayment: slots blocked for consultations that were never honoured.
- No qualification form and no automated follow-up. Every new matter started from a blank page.