114 comments

[ 3.5 ms ] story [ 1241 ms ] thread
I admire what you have done, but for a luxury experience, I do not want to talk to an AI that just tells me what is already on the website. If I have gotten to the point where I am calling you, its because I couldn’t find an answer to my question on the website in the first place.
No idea what `luxury` is doing here, but if I get an LLM receptionist, that ain't it.

This isn't to disparage the project - I think this sort of usage will become very common and a decent standard that produces good consumer surplus in terms of reduced costs etc. Especially impressive is that it's a DIY family-first implementation that seems to be working. It's great hacker work.

But be warned it will erode - in general - the luxury previously associated with your brand, and also turn some customers away entirely.

Is RAG even necessary here? Minimal information like a couple of price list with job times and opening hours should easily fit into any context window, right? It's not like he's dumping entire service manuals into the vector database here...
Yeah, this architecture is completely unnecessary.
It most likely isn't, but it seems like this project was more for learning purposes than for anything else. In that case, why not go for the "production-ready", "highly scalable" solution? I sometimes do the same for my personal projects. I over-architect them not because it's necessary, but because I want to get my hands dirty and learn something new.
Yeah, just stuff the entire website and pricing table into the context window.
For voice conversations the issue can be more latency than filling the context. Without knowing the site is hard to say, but if he had multiple pages worth of text (dunno, type of cars, procedures, some emotional story, etc.) and a "slower" model, it might be worth it to use RAG to preselect fast a small portion and use LLM to refine the answer.
I understand the other comments in this post, I too would be allergic to this sort of experience - luxury or not.

However, does the regular "joe/jane" feel the same way? I imagine my mom or dad would most likely not notice or care if they did.

"No hallucinations allowed" :')
That made me laugh a bit as well. Definitely want to see some rigorous testing on that, I'd expect that on longer calls tha caller can make the ai say basically anything.
>and he’s losing thousands of dollars per month because he misses hundreds of calls per week. He’s under the hood all day. The phone rings, he can’t answer, the customer hangs up and calls someone else. That’s a lost job — sometimes a $450 brake service, sometimes a $2,000 engine repair — just gone because no one picked up.

How much does it cost to have an outsourced receptionist? Even if it is 500 a month if we are really talking about thousands of dollars per month lost your ROI is still crazy.

Never mind an outsourced receptionist, some of those calls could be handled simply by the mailbox. Of course, some people will hang up once the mailbox message starts - but then again, some will also hang up once they realize they're talking to an AI chatbot, so...
> That’s a lost job — sometimes a $450 brake service, sometimes a $2,000 engine repair

Christ just hire some local teenager or whomever. There's people who will work for minimum wage.

The ROI is whatever the blog is advertising for -- AI training courses and such.
Ignore the expected negativity, many here have not used the latest gen of voice agents in development. Even if used as a router , prefer that to waiting to get through
Honestly great work, but this is very much something where the results matter more than the product. It ends without a single comment about whether it worked in Production.
Why not gpt voice directly instead of elevenlabs for voice and sonnet for intelligence?
This is an LLM generated slop post.
> Wired up Claude for response generation — The retrieved documents get passed as context to Anthropic Claude (claude-sonnet-4-6) along with a strict system prompt: answer only from the knowledge base, keep responses short and conversational, and if you don’t know — say so and offer to take a message. No hallucinations allowed.

Claude will hallucinate anyway, sometimes.

I don't think there's any way around this other than a cli or MCP that says "press the 'play prerecorded .WAV file button that says the brake repair service info and prices.'"

At the moment I'm pretty inclined to hang up if I feel I'm wasting my time with a robot.

But maybe soon we will not even realise we speak to a robot, given the current speed of ai development.

I wonder how that will erode trust in calls. I moved from cold emailing and cold LinkedIn to cold calling because of the massive amounts of ai spam I have to compete with. But maybe cold calling will die soon as well if the robots emerge.

You would probably hang up if it went to voicemail too so the net loss is 0
I've been car shopping recently, and it took me a full week to realize that every dealership I'd talked to had an LLM with a fake name handling customer intake. I was 4 emails deep with one before I stopped to think about how plausible their near-instant response times were.
Yup. After painstakingly having the AI misunderstand what you're trying to say, you're connected to a human and you need to do your story again.
The amount of negative comments here to someone building something is incredible.

I appreciated your post and have some takeaways around text formatting for TTS in my own projects. Thanks!

The site is selling SaaS templates and AI courses. The post is just an ad for whatever services it is offering...
As with any AI post lol
Definitely very interesting to me too. It was nice to see the breakdown of what technology was used and for what. The whole breakdown makes me believe I could build something like this too
Everytime I read or hear "the brain", my brain instantly shuts off.
I think we can solve this as a society by just making it clear that if you put an AI between you and your customers, you are absolutely bound by anything it offers them.
Think this through.

Assisted suicide? Handjobs?

Is Walmart bound by anything an employee says? Should it be?

Not a single clip/recording of how this sounds?

Like CMON this is the bare minimum here.

If the mechanic is under the hood all day, sounds like business is well and he can't support any more customers. Time to increase rates.
Maybe I am in the minority here, but I appreciate the new crop of LLM based phone assistants. I recently switched to mint mobile and needed to do something that wasn't possible in their app. The LLM answered the call immediately, was able to understand me in natural conversation, and solved my problem. I was off the call in less than a minute. In the past I would have been on hold for 15-20 minutes and possibly had a support agent who didn't know how to solve my problem.
When the problem is well-defined, the backend systems are integrated, and the AI has actual authority to act, it can be dramatically better than traditional support queues
I wish all shops just have a clear email address. Id much prefer emailing over placing a voice call...
(comment deleted)