benmorris
Joined 13 years ago
Joined 13 years ago
I'm taking that information and building a voice version. I'm taking a lot of time to make sure the phone version works naturally and that is can address most common things people call in about. Just like my chat version, GPT 4o was barely good enough to do what I needed. Since GPT 5 and on it has been more than adequate. I expect the same thing over time with voice, Open AI is going to release the GPT-live to the API which is a noticeable improvement.
Some people are just offended at the idea of talking to an AI chatbot voice or phone, no matter how good it is. I don't know how good this Kinney Drugs bot was, but it's very easy to make something that works but ends up being very frustrating for customers.
https://developer.paypal.com/braintree/articles/risk-and-sec...
I’ve been a PayPal/Braintree merchant for over 10 years, and this feels like a pretty big shift in risk.
For anyone not deep into disputes:
- A customer files a chargeback. - The merchant can submit evidence and may win that first round. - The customer can then file a second dispute, which goes into pre-arbitration. The merchant can again submit evidence and, historically, sometimes win.
If either side pushes further, it goes to arbitration, where there’s usually a few-hundred-dollar fee for the losing side.
Under this new policy, for transactions under $1,000, Braintree will automatically accept the pre-arbitration in favor of the customer. There’s no second chance to present your side as the merchant; the dispute is simply closed and refunded to the cardholder.
Practically, this means:
- A customer can lose the first dispute, - Immediately escalate, - And automatically win the pre-arbitration if the transaction is under $1,000.
I’ve already had multiple cases where bad-faith customers was awarded the full amount at pre-arbitration solely because of this rule. At that point, the only remaining recourse is to pursue the customer directly (legal action, collections, etc.), which is usually not realistic for sub-$1k orders.
For anyone running ecommerce on Braintree, this effectively creates a “free second shot” at a dispute for customers under $1k, with the merchant guaranteed to lose the second round by policy.
Has anyone else run into this yet?
- Mobile Eye dumps Tesla in AP1 days forcing them to develop their own vision based auto pilot aka AP2 https://arstechnica.com/cars/2016/09/tesla-dropped-by-mobile... For those that aren't aware it took years for it to work nearly as well as AP1. - Tesla removes rain senor in favor of vision based detection in AP2 cars as well. To this day vision based rain sensing is one of the worst features of the car. It is laughably bad in certain scenarios (night). - Early 2021 Tesla removes radar in favor of vision based cruise control and Traffic aware cruise control. Again, the product was rushed out due to supply shortages in an unfinished state. Still to this day the speed limit and distance settings are not on parity with radar enable cars. Vision based cars still deal with frequent phantom braking (just google it). Surprise surprise this one was so bad radar is coming back in hardware 4.
I've AP1,2, and 3 cars now and we were forced to sell our 2021 Y over issue #3 the phantom braking in the vision based system. I've heard it has gotten better but this parking sensor business is just more of the same from Tesla. Another one that is flying under the radar with HW4, Tesla is removing the ambient outdoor temperature sensor. This is right up there with the rain sensor. I can't tell you how many times I've been traveling through a weather system or going up and down elevation it's nice to know the actual temperature, not something fed from an API.
However,the process sucks. These days if a customer wants they can essentially get whatever they buy for free. I've gone down the rabbit hole several times leading up to arbitration and given up. In most cases it is a few hundred $ so it just isn't worth the fight. I 'win' the first round most of the time, the issue becomes if there is a second dispute for the same transaction, that is when it becomes nearly impossible to win without more compelling evidence.
I've had chatgpt help me with the following just this week among other things.
- solved a complex issue with scaling and transformations using pixijs upon providing a code example. The solution was 99% there and saved me hours of debugging.
- refactored JavaScript to the async/await pattern
- simplified a complex nested logic in c# with citations
- scaffolded a MySQL create table statement upon providing several c# classes. Again saved me the tedious steps of typing them out.