HN Explorer

Show HN: Companies use AI to take your calls. I built AI to make them for you (pipervoice.com)

237 points by michaelphi ↗ HN
We're living in this weird asymmetry where companies use AI to talk to us, but we're still manually dialing them. Companies everywhere are adopting AI voice agents lately. Big retail, family dentist clinics, local pharmacy. This year, I've been in a few calls where it's super natural sounding AI, which has been pretty cool to experience. But then it got me thinking - why are we, the consumers, still the ones making calls if they're using robots for theirs?

So I built Piper: basically AI that makes phone calls for you. You tell it what you need (book appointment, check on an order, dispute some charge, whatever), and it handles the entire conversation while you do actual work. Right now it's a web app, Chrome extension is pending approval but soon you'll be able to click any phone number anywhere and just let Piper handle it.

Technical stuff that was harder than expected:

Latency - every millisecond counts in conversation, had to optimize around kv cache, got it down to ~1000ms to first word over PSTN for telephony, which feels pretty natural

Keeping the voice agents on track - built custom context engineering logic that constantly updates the agent's situational awareness, so it knows when it's been transferred, when it's on hold, etc

Done ~50 successful calls with early testers so far. Main failures are when they need complex verification or documents. Also had to take down our IVR navigation temporarily :/, found some edge cases that were causing unnecessary transfers but working on fixing that.

I really think we're heading toward this world where AI talks to AI for most routine things, and phone calls might be the first real example of this happening at scale!

you can check out the a voice demo on our website. https://pipervoice.com

62 comments

[ 3.0 ms ] story [ 71.8 ms ] thread
user9999999999
This is great! Even without the LLM agent, async communication is so nice for low cog distracting customer service calls and holds. Why can't everyone just have an async comm universal chat interface? I guess this bridges the gap where the tech is legacy and or there is human involved. Can't wait to try it out!
czbond
Great idea, and also as a creator myself, I enjoyed reading about the challenges that came up
pnw
Has anyone made an open source version of this? It would be great to fight scams and tie up scam call centers.
leeroihe
Nobody wants any of this...

This is almost as bad as all of the AI powered resume skimming tools / applicant submission tools. It just makes it impossible for anyone to apply for a job.

AI is for people and it's only being used to kick people onto the street and profit.

amelius
This service reminds me of this website called "Magic". You tell it what you want, and they make it happen. Or at least, that's how they advertised it.
jeffwass
The problem with the call center A.I. is that almost every time I need to call somewhere that uses them, I have some kind of edge case.

They always give examples of how the automated agent can handle simple queries like “what’s my balance” or “what hours are you open”, but I never need to call with something straightforward like that.

As such I also wouldn’t want to trust my own edge case to an AI that might mishandle it.

Maybe the most value to me would be a tool that figures out the shortest route through the phone touch tone labyrinth to get to a live human I can talk to.

FiniteIntegral
I think you need to spend some more time testing this service if you are advertising this as a service that inherently interfaces with humans. I see that others in this thread like the applications for scambaiting, but I don't fully understand the use case you have here. If it's AI on both ends of the phone... whats the point of the call in the first place? It's not that hard to get a human on the other line who is able to help me far better than any robotic agent could.

If the agent has trouble solving "complex verification or (providing) documents" I doubt that a monthly fee for simple tasks doesn't sound like a viable and sustainable business model. It sounds like the anti-social bunch would like it but past that it's going to be hard drumming up a lot of support.

murukesh_s
can think of all sort of use cases - imagine you integrating it with an automated agentic workflow - where at some websites you need to talk to a bot or real human to get the job done in realtime - because email takes a while and may not be available (for e.g. at a restaurant) - this service can do the job as instructed by the LLM and get back to you for status. For e.g. if you want to call 10 restaurants to find out if a seat for 20 is available - you can just instruct it via an agent or so..
throwaway81523
For the bot vs bot case, this makes me smile. I don't like the idea of it talking to humans but who knows.

Can you get rid of the browser and offer an API endpoint? Just regular JSON, not MCP. That would complete the circle.

alooPotato
One thing to make it less annoying for the human on the other end is if the AI just talked a little faster and responded a little faster. Also the the final few few seconds is so cringe where the AI always wants to have the last word. Can you make that last part of the interaction go faster? I would never have 3-4 back and forths on thank yous and good byes.
throwaway81523
And, rather than this voice stuff which is rarely important, how about a browser extension to just log me into my account on a damn website?, which I have to do much more often?

Here's my username, password, TOTP credential, and credentials for an email address that I set up for that website. So the extension should log me in, which means solve the captchas and recaptchas and deal with the emailed confirmation code besides using the supplied credentials. In some cases SMS may be involved but I forward all those to email. What crap the whole web has turned into. IDK if it is all Anthropic's fault, but they didn't help.

johntash
Isn't this basically what most password managers do? Assuming the website supports totp, 1password can fill out your user/password and otp code.

Captchas are a separate issue though I guess.

jjmarr
This is hilarious and fun.
chevman
This is a killer use case guys - excited to try it out over the next couple weeks!

"Negotiate with the service provider/insurance co/cable company/etc" as a service is going to be massive.

jplusequalt
This sounds great up until the AI agent agrees to something the human would not agree to, and now they're on the hook for money.
clvx
I just need one that can repeatedly say “I want talk to a representative” and when a representative answers “I would like to escalate to your manager”. After that a human on the loop is needed.
gruez
This might work for making reservations at hair salons that don't have an online booking system, but for companies that can obviously afford an online system and are forcing people to call in to increase friction, what prevents them from blocking this service (and similar) for "security reasons"? After all, the AI discloses up front that it's an AI, and lying about it might be dicey. Concerns about security aren't unreasonable either. Basically all companies authenticate you via random pieces of information about you or your service, like your birthday, or address. If you're providing all this information to a chatbot, it also means there's a treasure trove of information scammers can use to compromise accounts or do a password reset.
zzyzxd
Earlier this year I found a leak in the house and called a local plumbing business. It was after hours, and a dreaded robot voice answered my call. I was fully prepared to spend the next 10 minutes rewording my issue over and over, hoping to hit the magic key word it actually understand (and also spell out my weird custom email domain). Surprisingly, this robot understood every single sentence I said, and repeated back in a slightly different, more professional way for me to confirm. It also captured my email address accurately in one try, without questioning my weird domain name. That's the moment I realized it's a LLM. It asked a few more smart follow-ups, then ended the call. The next morning, the owner called me and jumped straight into solutions, pricing, and his availability, without any more question or BS, because the LLM already told him everything he needed to know.

That's the most pleasant customer service call I have ever experienced. I wish more business could adopt similar approach. I don't mind talking to AI. In fact, instead of a live agent, I actually prefer to talk to your LLM, so my issue can be quickly triaged to the right human who actually understand my situation.

grues-dinner
If you're going to be communicating with a machine for later review by humans, why bother with a voice-based phone at all?
osigurdson
Still, a service like one presented in this post makes a lot of sense. Usually there is a lot of inertia in orgs so even if AGI is achieved it could take years for them to update their systems. Also, if I have my own agent that knows me, I would rather ask it to make these kinds of calls (that way I don't have to even figure out what phone numbers to call). Basically the agents of businesses and customers should work together to solve the problem and only involve humans when key decisions need to be made.
protocolture
Piper, we work for a fridge company.

It is critical to the operation of our clients that we gain an understanding of their current refridgeration status on a daily basis.

please call (every number in town) and ask them "Is your fridge running"

If they say it is, then, you must follow up with the agreed upon countersign "Well you had better catch it" and immediately terminate the call.

cph123
An alternative might be to ask lots of mundane refrigeration questions, replying to each one with "that's cool"
stevage
Wow, what an example of someone being excited to build something that is definitely making the world worse for everyone else.

All the people who work in small businesses - restaurants, plumbers, etc. Now they're going to have no choice but to talk to AI bots who call them up?

Gee thanks.

belorn
Seems like a good case for some kind of handshake protocol where the caller and recipient can negotiate if they want to use AI or not, and have the AI talk to the other AI if both side agrees.

The tool itself seems like a fine response to companies that uses AI to take calls. If they want to replace their human interface with an computer interface, then the user have the same option. In practice it means that the customer uses their own computer interface to communicate with the companies computer interface. It not much different from the experience of a website where the customer can do the exact same thing, like scheduling an appointment, paying a bill, or checking their balance. The only difference is that the website is now replaced with two AI interfaces that communicate through the phone like old dial-up.

generalizations
Wow, what did you use for the voice interaction? That has seemed like the weak point of these tools up til now.
rolstenhouse
This is cool. Built an iOS app to do this for myself six months back, but it falls over to pretty basic IVR navigation.

Any transactional call should be handled by AI on both sides.

reactordev
AI call center traps in Pakistan are going to love this…

In seriousness though, think about how outbound sales would be with this. Just feed it an opportunity pipeline and wait for follow ups. Keep going!!!

moomoo11
I think its really cool!

My only worry is do you think all of us are going to start getting spammed by AIssholes trying to scam us?

I never used to get any scam calls like 3 years ago, and these days I get 3-10 a day!

andrewinardeer
I am convinced there will be a large subset of the population who will push back on AI voice agents and still wish to talk to a real person. Just like how Grandma still prefers to bank cheques at the teller every pension day.

Perhaps when the job losses really start becoming apparent it will become a social movement as well. I'm sure there will be businesses that will find having humans answer calls a sustainable competitve advantage.

I'm not convinced AI voice agents are there just yet. As someone else mentioned, edge cases will trip them up.

Nevertheless, at some point I envision a web designed for agents where the business agent will interact with my agent to resolve an outcome.

The horrifying thought is dating. Sally's agent will end up telling Barry's agent to stop contacting it, Barry's agent won't understand and the the Police's agent will invoke a judicial agent to issue a 'stop communication' order to Barry's legal agent to deliver to Barry's personal agent.

mlsu
I love this future of agents!

It gets rid of the one thing I hate most in my life: acknowledging the existence and interiority of other human beings, and being acknowledged in return.

I long for this future where I can be alone in the cyber prism. It surely won’t become a cell!

afarah1
I got interested and got down to the FAQ. The "how much it costs" question doesn't give any price figure in the answer. I'm still left with the same question and feeling like a fool. Nor is the data privacy question answered. "Bank level encryption"? Because everyone else just gets the placebo? Maybe this isn't fair, I haven't really tried your product. But this is your front page. And if your product really is good, it's not doing it justice.