Don't pay for AI support failures, says Gradient Labs CEO

Dimitri Masin, CEO of Gradient Labs, argues that companies using AI agents for customer support should only pay when the bot does its job. "If you look at Salesforce, they price the automation per conversation," he told The Register in a phone interview. "So essentially, if you have a conversation with AI, no matter what it …

[Anonymous Coward]

Only pay when the bot does its job

What is the man doing? This is crazy talk! Is he trying to ruin the AI suppliers?

And how many fingers can we keep crossed that he will succeed?

[ChrisElvidge]

Re: Only pay when the bot does its job

I've said for years - only pay on successful implementation.

[Neil Barnes]

Re: Only pay when the bot does its job

It worked for traditional Chinese doctors: only get paid while the patient is healthy.

[Wang Cores]

Re: Only pay when the bot does its job

Boeing suicide special coming right up.

[Lusty]

You still pay the frontline support staff if they can’t resolve an issue. This is cheaper frontline staff. You still obviously need 2nd and 3rd line support.

[Anonymous Coward]

Let the support requestor push the pay button.

Has your support request been completed to your satisfaction?

⬆ Pay the Bot. ⬇Bloody useless, dock the blighter's pay.

[O'Reg Inalsin]

Frontline support is there to keep the customer from advancing

Wear the customers down with attrition. That that makes up half of all support activity, and that AI can do it too, is entirely believable :(

[druck]

Re: Frontline support is there to keep the customer from advancing

Pissing the user off with AI so much that they stop asking questions is counted as a resolution.

What dies the user do then? Ask an actual person they work with, thus wasting their time too, so how is that increased productivity looking?

[breakfast]

What do you let it do?

I know this topic is in the air at the moment, but when I read something like this I'm immediately wondering how much power the AI is going to get - the more it has, the more of your system is vulnerable when someone figures out how to exploit it.

"We configured the AI to be able to refund orders for some customer complaints and now our bank account is empty" kind of a beat.

[Tron]

Cripes. Google won't like this.

If AI throws the spotlight on the novel concept of only paying for something when it works, what would that mean for Google's business model, charging people for showing adverts?

[FuzzyTheBear]

Re: Cripes. Google won't like this.

Would be interresting that Google only gets paid when showing an ad to a person translates in a business transaction. They would probably make a hell of a lot less money.

[Claptrap314]

Re: Cripes. Google won't like this.

They know how much. At least they did at one point. Then they mostly forgot. But only mostly.

[JimmyPage]

The Peter Principle

Applies to systems too.