OpenAI observation shows Claude 5.5 Pro gaining professional adoption through deliberate packaging and vertical use-case marketing rather than raw model performance

There is going to be a point where AGI starts spreading like fire. I think it's sooner than later. Shit is going to go bananas

It's an open secret that AGI is here. When the rest of the populace actually really updates on this, I'm worried about the mass spiritual reaction. I have been doing 5 years (which is relatively late..) of serious processing. They are going to need to do it in an instant

Like, for the first time in a long time my in laws have started to take me seriously. When I was telling them about this stuff 5 years ago (!)

It's weirdly vindicating to hear others that I've been warning finally take me seriously. Like yep, I was right

But also concerning

Here is an example of what I mean. Right around the o1 public releases, I said that AGI was already here. People didn't really take it too seriously. It was along the lines of "alright, I'm calling it, it's AGI"

Guys. It's AGI. You haven't noticed?

I wasn't joking..

@tszzl Most professionals and management teams anchor their understanding of capabilities and reliability to MS Copilot's terrible offering. No surprise these people are AI skeptics.

I cover this dynamic in detail in my review of AI in the NZ public sector: https://tinyurl.com/NZ-AI-adoption

@tszzl Is the marketing strategy just packaging the prompts for each vertical well?

@tszzl bingo, end-users dont give a fuck about benchmarks. benchmarks are not concrete

all they wanna see is a problem they know deeply being solved in a concrete way. it gives them something they can point at and say "i want THAT"

@yacineMTB I have said since 2023 and it really happened it will be the same in 2026 but from many billionaires and millionaires they continue to push technology inhumanely but they know it is pushed with intelligence continuously especially with the united nations and european union which

@tszzl Physicians are even tougher to sell too. They like openevidence, which is free, but very mid.. I did convert several to chatgpt Plus though after they actually see what it can do

@tszzl I know an older lawyer that uses a typewriter and has his associates retype on the computer.

One of the things I often say to engineers is that it’s easy to build a product that serves many customer types and use cases—but surprisingly hard to market a product effectively to many customer types.

Engineers often think the limiting factor of serving multiple customers is engineering complexity/feature complexity, but actually I think it’s that marketing/brand is hard to straddle across many segments.

As AI makes companies do more and serve more customers than ever before, I think we are going to see companies have to split off into many “microbrands” in order to get into all crevices of the market

@scottastevenson @tszzl Yes, it’s like trying to be a restaurant that serves different types of cuisine. Even if it’s good, the perception is that a place that specializes must be better since their attention and efforts are not scattered.

I've used Claude and GPT for medical searches during shifts for a while; initially out of curiosity, now out of convenience. The answers fed to me were often about 50% correct, and obviously biased if you knew the underlying actual evidence.

I know use GPT as an assistant.

Big difference in actual utility, relative objectivity (I think), and ability to actually teach it what I would like in responses.

@tszzl Keep trying and tell me when it fails is one of the best ways to pressure test esp with domain experts

@tszzl You see that problem they pointed at? "It needs to stop making up everything"

THAT is what has made Claude successful, not the packaging. Professionals CANNOT HANDLE the inaccuracy, and the anxiety they may have missed something. You have to get it under control or it's unusable

@DrBeavisAI @tszzl Funny part is there’s this for clinicians as well

https://chatgpt.com/plans/clinicians/

@tszzl bro dont think i forgot about this

@tszzl I’m a lawyer (arguably elite, inarguably not your friend (although a fan)). Your Pro line models have long been the best at complicated legal tasks. It’s not a close question. The Claudes are delightful. They cannot navigate an intricate legal question the way 5.5 Pro can.

@tszzl 5.5 pro basically never hallucinates. It just doesn’t have accesss to most case law