Activity
Mon
Wed
Fri
Sun
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
What is this?
Less
More
3 contributions to Brendan's AI Community
I built a full AI receptionist with ONE prompt
I just tested how far one prompt can go. I gave Claude a single build prompt, and it set up an entire AI receptionist for an HVAC company: ✅ The voice agent (Vapi) ✅ The phone number ✅ The backend automations (Make) ✅ The owner alert email when a lead comes in ✅ The client dashboard No clicking through Vapi. No dragging modules in Make. Claude did it through the Vapi and Make MCPs. The test call: I played a freezing homeowner whose heat was out. The AI answered, collected the details, and the owner got an alert email in seconds. Why it matters for us: setups that used to take days now take one session. If you sell AI agents, your margin is no longer in the building. It's in picking the niche, writing a prompt that works, and getting it in front of clients. 🎥 Full build (test call at 6:38): https://youtu.be/VxJLJ7cOqOo 📘 I wrote up the exact prompt and step-by-step so you can copy it: https://venteeve.com/guide/hvac-receptionist Try it for your own niche (dental, plumbing, remodeling) and drop your result below. Tell me which niche you picked, and I'll help you tweak the prompt. 👇
0 likes • 13h
@Mouldi Nouri most definitely! It requires a lot of testing before it is fully production-ready for a real client. Which is usually the testing period. I just wanted to show that if you have the right set up, you can have most of the work done already within just a few minutes. 🙌
voice AI finally sounds like a person on the phone
just tested the new full duplex speech to speech model on my AI receptionist and the interruption handling alone is wild clip: https://youtube.com/shorts/6wk7j70dWVg?si=Bc9llSBp8XIwNOGd why it feels different from every voice bot you’ve tried: cascade bots chain three models: speech to text, LLM, text to speech. each step adds delay, and the bot has to guess when you’re done talking. that’s where the dead air and talking over you come from. turn-based speech to speech fixed a lot. one model, audio in and out, faster and more natural. but it still waits for you to finish before it speaks. cut in and it stumbles. full duplex doesn’t wait. it hears you while it’s talking, so a “wait, actually” or a quick “mhm” changes what it does in real time. nobody is waiting for a turn to end. that matters most for receptionists, because real callers ramble, interrupt, and change their minds mid-sentence. anyone else building on this yet? what are you using it for?
0 likes • 21d
@Brendan Jowett Thanks, glad it landed! Trailing off vs jumping in is honestly the trickiest part with full duplex, since it's listening while it talks. It's picking up on whether the sentence actually sounds finished, not just whether you paused, so an "I was wondering if..." doesn't get cut off. A quick "mhm" doesn't stop it either, but a real "wait, actually" does. It's still not perfect, so I keep testing with messy calls and checking transcripts. Are you building with it too?
how are you showing clients what their voice agent is doing?
Hey all, this was one of my first real headaches running an AI agency. you build the voice agent, it’s taking calls, and then the client asks “so... how is it doing?” and you’ve got nothing clean to show them. my answer was a google sheet and a pile of automations to keep it filled in. every new client meant rebuilding it, and I was always waiting for something to break and a client to message me about it 😅 so I built a dashboard to fix it for myself. one API key and clients get their own login with calls, transcripts, monthly ROI reports and AI analytics. works with ElevenLabs, Retell and Vapi. I’ve since opened it up to other agency owners. quick demo: https://www.youtube.com/watch?v=O6wOmI7aNG0 curious though, what are you all doing for this right now? still sheets and screenshots like I was? Justine
0 likes • Oct 17
@Mouldi Nouri You’re right, usage billing is part of it 🙏 The app can invoice clients for setup, a monthly fee and per-minute usage, so they see exactly what they paid. The ROI report then shows what the agent made next to the monthly fee, like “4x what you paid”.
0 likes • Oct 17
@Brendan Jowett Thanks! Mostly outcomes and conversion. Clients get a weekly AI brief based on bookings, sales and what callers asked about, and they can ask questions about their calls in plain English. Sentiment only shows up when the platform provides it.
1-3 of 3
Justin Samson
3
13 points to level up
@justin-samson-5162
I run an AI automation agency that supercharges business growth.

Online now
Joined Oct 17, 2026
Powered by