Hacker News Reader

I had to go to an xfinity store the other day, and seeing the things people come in for made me realize why AI is attractive to companies. The four or five people in front of me did not need a human in the loop for their issue. If these people could go to xfinity.com and ask some bot where they can find their bill, how much they owe, or if their internet is down, xfinity employees could focus on actually selling things. I imagine it's basically the same for every customer service.

> If these people could go to xfinity.com and ask some bot… if their internet is down…

See, this is how we get such useless chatbots.

I do this with cox all the time… because I also have cellular data plan

Though cox will perpetually try to gaslight me so it doesn’t help anyways

  > Agents (if implemented well) are an order of magnitude more effective at resolving issues compared to a call centre worker who is reading off a script and churn within 9 months

For this to be true, the agent needs to actually be given the means to solve the problem, otherwise an "agent" is just a glorified help page that wastes your time.

But it seems like companies don't want to do this part, possibly because of fears that someone will trick the agent into giving them a refund or something. Or because the actual goal is to optimize for fewer costly refunds/cancellations/policy exceptions etc.

So for whatever reason, they stay stuck in that useless local maxima while simultaneously making traditional help increasingly difficult to get ahold of when needed for an overall net worse experience as a customer.

Voice agents have capabilities and policy to alter customer state. Just the other day I called into a CC company and the AI waived an interest charge.

Not all companies do that.

There's a certain vendor that requires me to place same-day orders by a specific time. You can easily place an order from the website. If you need to cancel one, you have to ignore the grayed out cancel button and call their cancellation support line. There you'll talk to an agent that doesn't have access to cancel orders, so you have to convince it that it can't help you before you can transfer to a real employee with the ability to hit the "cancel" button.

and just today i talked to a bot about a missing item from an order and it had to call in a rep to push the button to ship me the replacement. except the rep’s messages seemed to filter through ai as well so what should have taken 20 seconds took 2m between messages. it could be good, but as the other commenter said some places are in a weird shittier hybrid model.

From what I've seen, it's the opposite -- the whole value proposition of these companies is to take on brand liability and allow the agents to autonomously take actions.

> possibly because of fears that someone will trick the agent into giving them a refund or something.

Refunds could require approval. And, it could not be just the agent's sole decision.

I think this is going to be wonderful. I'll have my offensive AI call their support AI and prompt-inject my way to a rebate tier that nobody knew existed and nobody can cancel it because all the remaining humans have been reduced to phone-to-screen input machinery.

right? its practically begging for it to be tried -_-. i wonder if someone somewhere will turn a sim farm on such companies to try and mass inject them to do weird shit or say nasty things to other customers etc. - ofc youd hope its set up in a way u cant, but then again we learn yesterday all ur stored passwords in edge are in plaintext mem... i would not be surprised if some of these companies get totally crapped on by some adversaries or malicious parties.

People are not protesting hypothetical proper LLM tech-support, which indeed can be ok and cheaper than humans. People are protesting actual practical implementation of the LLM tech-support which they already experienced themselves, no need for second-hand retellings or stories/ads. In practice I had LLM of my goddamn bank (where I'm a premium and old client) hang up on me with a response "I don't understand you" and cut the call. And now I need to call them again, wade through a digital labyrinth again and wait on line again. Awesome. Or when I urgently needed help with a government ID application and the only official tech-support is an LLM chat which has approximately 20 super dumb scenarios explained and literally nothing else. And the only LLM sign is that now I need to type my query in free style and not select predefined buttons, but the result is the same. So I had to resort to going to Facebook (thank St. Mark for this "innovation") and beg for human help in the promo page of that application (and I got human help there, lol, don't delete your FB accs people). Or when my internet got cut (cable line fault was discovered later by a technician) and LLM of my MSO fucking banned me, because their system was bugged and kept disconnecting me from their end and I exceeded a really small number of retries (like 8?).

I'm pretty sure every one commenting here has their own horror story about LLM support. Now that is what people are angry at.

SWEs get paid to get good at reading documentation on processes. I think HN is biased since we'll only escalate once documentation can't help us.

I'm also bullish because AI coding agents give up easily if my problem is complicated.

I think it'll be easier to convince an AI to transfer me to level 2 support than a human.

> It's always interesting seeing how HN reacts to AI CX (as someone who works in this space). Yes, the tech savvy crowd loves to say how they always ask for a human and love old school phone trees in reality 50-80% of callers come in with easily answerable questions because they don't know how to nav the website and prefer to ask in natural language

For the other 20-50% it's much worse, and that's the problem. And people on HN will tend not just to fall in that group but in the top 5% of "least solvable by just reading the website" questions.

Yup.

Raises the floor at the expense of lowering the ceiling.

And another thing...

Sierra says:

> Transform phone support with AI agents that speak naturally, reinforce your brand, and take action—across inbound and outbound calls.

Reinforce one's branding? For the better? Really?

Seems unlikely.

I've LARPed in most roles around product development. Tech support, sales, QA/Test, tech writing, marketing, etc. Enough to appreciate that engineering the entire lifecycle is important.

A comment elsethread states customers HATE these support robots. I believe it; me too.

Before adopting agentic CX (of whatever its called), I'd worry about alienating my current and future customers.

Penny wise, pound foolish.

At the risk of being labeled as racist, I'll take an LLM chat bot either in text or delayed voice to an outsourced Indian call center any day. This isn't an indictment of Indian's and their ability to communicate. But the type of folk Indian call centers tend to fill their worker pool with to keep costs adequately low. I've worked with a ton of amazing folk from India, but they are not the lowest common denominator that call centers tend to hire from.

I have yet to encounter an AI agent that was able to handle my support questions adequately. I always end up having to get a human (which is becoming increasingly difficult or virtually impossible).

I'm sure AI Support Agents will be implemented better, but so far in my experience, the humans I connect to far outperform the AI agents.

that's fair, most implementations in the industry are in the early stages and implementing a full powered agent with access to all the tools it needs is hard (very political as you can imagine). i hope over the next year you notice them getting better!

the thing is, I don't care if the AI agents get better; I want to speak to a human who has the cognitive ability, flexibility and authority to handle my problem.

I understand some people call with trivial or easy questions, and those might be handled just fine by an AI agent (just as they would by a human). But if I'm calling it's because it's not easy or trivial, otherwise I would have figured it out. Calling is always a last resort. And it's also because I want something handled quickly and don't want to spend time trying to navigate a maze of questions to try to get the AI agent up to speed and do what I want. So all the agent does is make me upset and not want to do business with that company again. And the more difficult the agent makes it to get to speak with a human, the more unhappy I become.

Stop trying to cut costs and extract maximum value while making things worse for customers, and stop trying to tell yourself -- and us -- that this somehow provides a superior experience for customers. That's BS. I've seen the step by step decline in phone customer service over the past 30 years, it has clearly never been about the user experience.

thanks for your insights, however, citation needed that they will get better

An AI customer service bot told me my autopays dont come out because they're scheduled for the 31st, and not every month has a 31st day.

I review recordings from calls routed to Sierra and a few other similar systems on a regular basis for <day job>. The calls come from folks of all walks of life, not just tech folks.

I’d say the vast majority of callers absolutely hate talking to these things and spend most of the call trying to get to a human, often getting frustrated and hanging up (shows up positive in the metrics, call handled without transfer!).

Though I’m not sure the companies deploying them really care, they’re just happy they can fire call center employees.

Tbf I've thought a decent bit about how most current AI is essentially just being used to digest what exists on a website/etc. Honestly even just the vector search/RAG part is useful, but more-so with a model to help do some initial filtering of it.

It's an odd use case - we have used language for a millionish years or so and it makes sense that that's the easiest way for us to get at information/do things.

But at the same time it's faster for me to read than listen, but it's often slower to type than to speak. It's faster to hit one button in a familiar place to do some predetermined thing, but much slower when the location of that button changes/gets hidden under submenus/I'm not familiar with an app or website.

On Android I constantly use the search function of the settings menu and I feel like this will be the golden UX going forward - a side by side UX + NL interface. So I can ask "how do I add a photo" and from there I get taken to the right place and can continue to add multiple photos in one go following the same pattern.

Though I suppose the nicer alternative is just "add all the photos I took near the waterfall from today".

The problem is their bots try to get me to input what I need only to reject/get confused af what I write and give me super limited options or the classic runaround. I can’t tap my way to the solution. I am used to menus, I am used to proper UI’s. I don’t know what language each company uses and apparently their crappy reskinned Gemini bots can’t translate regular speak to it. But if I can see the words and see what leads where, I can figure it out quickly rather than expecting a facsimile of a real person to play middleman between me and the phone tree. It’s basically just navigating it and occasionally skipping a step or two for me. The loops I get thrown in to are such a con it’s not worth it.

I went through this whole song and dance the other day with Uber. I needed to change something and the “AI helper” kept trying to force me into the lost item tree. They snipe keywords and ignore everything else. If you say “reservation” or “cancel” that’s all it works with with none of the context.

Making me talk to a fucking robot leaves me with a deep and abiding hatred for your company. I will prefer almost any alternative to doing business with you and hope fervently to read about your bankruptcy.

What percentage of interactions having this result will cancel out your cost savings?

Somehow, I think we're missing the point and maybe braincells are being sent in the wrong direction. Well designed products don't need good customer support. My toaster works well. Haven't called them once.

If we are designing a thing is so terrible that it makes customer support necessary (other than the obvious corner cases that ai cannot solve) then sure, let a computer do it. We’ve already failed at every other step.

Your toaster is well... a toaster.

And even something as simple toaster might need some customer support. These things do fail quite often, sometimes dangerously.

Increase complexity even a little; or worse, deal with a service (phone, internet, subscriptions, etc) and you will need this safety net.

Nothing is 100% reliable in this world.

There has never ever been a case that AI has resolved my query, except the simple decision tree for things like refund. Have you got any positive experience with gen AI in any of the site?

Cant wait to spend my day arguing with a phone support clanker about why my medical insurance claim was rejected by the medical claims clanker, only to get forwarded to the 2nd tier "patient advocate" clanker who's really just the medical claims clanker in disguise.

The future is extremely dystopian and sad right now. The corporations are not going to use this the way you think they are. They are going to use it to maximize their profits, not help their customers.