Blog
What we measured, what we changed our minds about, and what we tell clients when the honest answer is no.
Case study: a phone agent that books the appointment and files the paperwork
Most phone agents read a script and hand you a transcript. This one identifies the caller, answers from live stock data, books the slot, and writes the contact, the deal and the call summary back into HubSpot before the caller has hung up.
- voice agents
- CRM
- case study
- measurement
Case study: a WhatsApp number that answers from your documents and asks before it acts
Forward it a contract and it becomes searchable. Ask what is on the calendar and it tells you. Tell it to email a client and it shows you the message and waits. Running in production, with the approval gate enforced in the database rather than in a prompt.
- assistants
- retrieval
- case study
- trust
Can you trust an AI assistant with your company email?
Not unconditionally, and any vendor who says otherwise has not thought about it. The safe version is a design decision, not a better model.
- assistants
- safety
- design
How long should a customer wait for an AI phone agent?
We measured it on a real deployment. The answer is not the number most vendors quote, and the thing that fixes it is not the thing most people try.
- voice agents
- latency
- measurement