Why contractor calls are different
A homeowner with no heat does not leave a voicemail; they call the next company. A landlord with a leaking water heater wants a time window, not a callback. And the calls arrive when the crew is on jobs and the office is closed: early morning, evenings, weekends, and the first hot week of summer. A human answering service takes a message; an answering service for contractors has to decide whether the call is an emergency, dispatch it if it is, book it if it is not, and put the job into the software the dispatcher runs the day from.
This service is built around that decision. Your emergency criteria, your service area, your job types, and your pricing rules become the agent's script, and your field service software is where every call ends up.
What the caller experiences
The phone is answered on the first ring with your company name. The agent asks what is going on, where, and since when. If the answers match an emergency rule you defined, the caller is transferred to the on-call technician while the agent sends the address and symptoms ahead; if the technician does not answer, the backup rule runs. If it is a routine repair or a new system estimate, the agent checks the address against your service area, collects the equipment details and access instructions, offers the next available window that fits your zones and technicians, books it, and sends a confirmation text. Commercial and property-management callers get the extra questions you want asked, such as unit numbers and billing contacts.
Callers who ask for a price hear only what you approved. Anyone the agent cannot help by rule gets a message path or a transfer, never a guess.
Built on your number and your dispatch software
DPI runs the agent on an Asterisk-based telephony core connected to your phone system or your carrier over SIP; your numbers, including ad tracking numbers, forward to it in the mode you choose, and it can transfer to any mobile or extension. Jobs, customers, and appointments are created in ServiceTitan, Housecall Pro, Jobber, or FieldEdge through their APIs, so the dispatcher sees the booking on the board the same way they would if the office had taken the call.
Speech and language models are chosen per workload: open-source speech models keep per-minute costs down at volume, and the language model can be hosted or self-hosted. Every call is transcribed and logged with its source, outcome, and the job it created.
What you will see
A dashboard with every call, its category, the jobs and estimates booked, the emergencies dispatched, and the transfers made, plus transcripts you can review. Weekly reviews in the first month refine the emergency rules and the questions the script did not anticipate, often by season. Monthly, we report answered-call rate, booked jobs, after-hours dispatches, and the changes shipped. As you grow, the same agent can send maintenance reminders, follow up on open estimates, request reviews after completed jobs, and cover additional locations or brands.