Activity
Mon
Wed
Fri
Sun
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
What is this?
Less
More

Memberships

Voice AI HQ

119 members โ€ข Free

๐ŸŽ™๏ธ Voice AI Bootcamp

5.7k members โ€ข Free

Nexis AI w/Jason

442 members โ€ข Free

Assistable.ai

2.6k members โ€ข Free

2 contributions to Voice AI HQ
๐Ÿค”Need sum help guys - VAPI vs Retell for my situation??!
Hey fellas, Iโ€™ve struggled in the past with Voice AI quality and reliability for my momโ€™s hair salon (I almost wanted to punch my agent but they're not real so I cant). I was using Assistable (Voice AI platform built on Retell for HighLevel users), but the results were disappointing and I was considering hiring a VA appointment setter from Colombia this past week because I genuinely thought the state of current LLMs just wasnโ€™t there yet in terms of quality. Then I watched Hugoโ€™s video on context engineering like 83 minutes ago and realized maybe the issue was my setup โ€” I was using a static system prompt instead of dynamic turn-by-turn prompting. Now I wanna take a risk and give Voice AI another shot. My goal is to build an agent (or agents) for the salon to handle appointment setting, reschedules, confirmations, FAQs, and objections. Iโ€™m treating the salon as a low-pressure experimentation sandbox to prove I can build a production-quality Voice AI, since I plan to launch a B2B collections firm soon and this is my chance to validate if I got motion or if I'm cooked. My tech background: - Very comfortable in HighLevel (this is my main CRM), Make, and n8n. - Can handle HTTP requests, data transforms, and light JS in n8n (I vibe code it tho). - Did I mention vibe coding? My proposed setup: Voice AI platform with MCP โ†’ send latest message to n8n โ†’ query vector DB โ†’ return relevant chunks โ†’ feed back into MCP. Context engineering achieved โœ…๐Ÿ•บ My main question: Given my background as a โ€œcitizen developerโ€/vibe coder (not a real dev), is it realistic to go straight into VAPI for more granular control (since thatโ€™s what Iโ€™d likely use long-term for B2B collections)? Or is the learning curve too steep, and should I stick with Retell for now to get something working faster? The goal is to get an agent up and running in ideally 5 days or less. Appreciate any advice from any grand masters and Voice AI wizards. ๐Ÿ˜ Anyways enough yapping, I will be checking for gold nuggets in the comments ๐Ÿ‘
0 likes โ€ข 16d
@Hugo Podworski Hey Hugo, thanks for the reply. I am leaning probably more towards Retell for now, I learned the hard way that whenever I estimate the deadline of a project, it's usually 2x my estimate. Plus I have a million other fires to put out because I'm wearing the media buying hat, sales hat, dev hat, etc. and Im not geting paid enough to be this stressed out. That being said, I also wanted your take on this, because even though I want to give Voice AI another shot, I'm lowkey still on the fence about this because production quality Voice AI OR a real human setter is about 2-3 weeks of time investment regardless of the path I choose, hence why im struggling to decide. After taking appointment setting calls myself for a week, I realize that there are nuances to the sales aspect that may be very difficult to replicate with AI such as - my ICP likes to yap sometimes, and the AI would have to be really good at bringing them back on topic or knowing when to "politely interrupt" rather than letting the prospect go off the rails - I can sense hesitation in people's tone and I know when they are bullshitting or about to make up some wack excuse, sometimes I go out of my way to pre-empt objections, or I may even cut people off and hit them with a reframe to take control of the conversation, and the thing is with Voice AI, it usually always waits for someone to finish, and it never prematurely cuts people off intentionally - I had many issues with Voice AI being too sensitive and stopping mid sentence anytime it hears a noise, making the conversation very awkward as there are long pauses (hopefully sensitivity can be reduced in Retell, idk of thats a feature) - I'm in southern california, probably 40% of leads are not native english speakers, so accents can cause difficulty with transcription, although I can also just use a spanish agent - I plan to leave the systems behind for the salon as I won't be doing this after new years. A VA that knows how to use the CRM, my database, etc. is critical cuz my 55 y/o mom don't even know how to copy and paste. (although I guess even if I use voice AI maybe I can hire a systems admin VA or something)
What Video Do You Need Right Now?
I'm knee-deep in planning the next few pieces of content and honestly, I'm tired of guessing what will be valuable for my audience. Everyone's throwing around the same recycled voice AI videos. So here's what I want to know - where are you right now? Are you still figuring out if this AI thing is legit, or are you already building for companies but failing in production? And what's the one thing that would actually move the needle for you? Not some theoretical framework - the specific problem that's keeping you from the next level. Most requested topic gets made first. But I need details, not generic requests. The stuff that actually works doesn't sound pretty in a title.
0 likes โ€ข 17d
What up hugo, please show us how to do context engineering through MCP servers. The video where I saw you showed the setup was the debt collection Voice AI agent where you were streaming LLM calls back to VAPI but that involve some sort of custom code middleware layer. i think MCP simplifies this, VAPI and Retell both have native ones now.
0 likes โ€ข 17d
Also it would be great if you could discuss the different options in terms of Vector DBs for RAG, and which ones you recommend we use based on our situation. Appreciate all the vids though you're like one of the most educated people in the space. so many shilling meatheads in the Ai space in general.
1-2 of 2
Ethan Khou
1
4points to level up
@ethan-khou-4293
Ethan

Active 4d ago
Joined Sep 10, 2025