🇺🇸 Building. prev. userogue (acq)
Jacksonville, FL
- Tweets 11,424
- Following 1,117
- Followers 3,071
- Likes 55,931
roast away!
personally i only care about building the best product for our users. whether its our model or someone else’s doesn’t really bother me. this is my own opinion of course
but i do hope that our next models will be competitive. let’s see what happens
88
18
1,473
53,268
My biggest q is how much usage will $300/mo get me on @bot if it’s running opus 5.5?
If it burns down way faster than grok 4.7 seems like a big L (unless there’s a way to toggle models easily)
1
1
200
John Shahawy retweeted
You’re living in a bubble. You need to understand that a whole world lives beyond the constraints you have manufactured or imagined.
For example the codex harness provides very little in terms of actual functionality - the same goes for most coding harnesses. They will be generally superseded by more capable specialized harnesses. Even if you scoped codex to only a coding-focused harness - which it really isn’t - it is lacking compared to purpose built coding harnesses that compete with it.
Right now labs have the advantage and can somewhat dictate terms but you should not believe that will be true forever. There’s seemingly a belief that we should be thankful for the (limited) capabilities we’ve been given- that we should allow a few companies in the world to dictate what we can and can’t do with technology. That’s never worked before and nothing makes this tech any more defensible.
In five years we will almost certainly be consuming local models for a majority of daily tasks, and for the things that can’t be done locally it’s going to be entirely commoditized. Once that’s true it means vendors will have to start competing by giving up more to the customer rather than the increasing restrictions being done today.
15
31
398
72,143
John Shahawy retweeted
Nvidia laid off nearly 70% of its staff. Nearly went bankrupt at least two other times. 90% market cap drawdown in 2001 that it took 14 years to recover from.
So much of business is making incremental progress over years and going all in when you see the wave.
2
15
286
9,599
If your company is structurally betting against ai progress…
Now is the time to adjust fire.
Your customers are not stupid.
Your product is not as sticky as you think.
Especially when your customer can have their new AI systematically replace you.
The short term solution is excellence and velocity.
The medium-long term solution is to pivot your business model to something that gets STRONGER as AI gets more powerful.
1
1
175
been working on something fun. can't wait to share. it feels like 2023 right before chatgpt dropped all over again. blue ocean
5
197
If AI makes your life feel easy/boring/not challenging:
That’s your sign to do something more ambitious.
2
218
John Shahawy retweeted
Rogue Fitness just announced the near completion of their newly expanded US manufacturing facility for squat racks, barbells, and more in Columbus, Ohio.
This puts their factory size at over 1,000,000 sq. ft. and the ability to process up to 250K lbs of steel per day.
This makes Rogue the largest single-site strength equipment manufacturer in the United States.
The most interesting part, though, is that Rogue is owned 100% by Bill & Caity Henniger who started it out of their garage.
60
53
1,294
184,065
this is a good framing especially in enterprise sales.
it's the difference between a paid pilot (you might not even realize its a paid pilot) vs a customer who is planning on renewing next year
i usually tell YC founders to get customers to pay ASAP, since that's the best signal that someone wants your product.
but sometimes the money is a much smaller ask than everything else that comes with using the product.
e.g. imagine selling a new tool to a sales team. the manager has the budget and likes the demo. getting a paid pilot might be pretty straightforward.
then somebody has to move the customer data over. everyone has to learn the new system. the manager has to convince the team it's worth being slower for a while as they figure it out.
or take a task management app for consumers: the actual test is whether someone remembers to put all their tasks there and check it instead of the notes app they already use.
we were discussing a version of this in office hours last week, and i think it's useful to ask: even if the product were free, what would the customer still be reluctant to do?
a payment is still great evidence of a need, but seeing your users overcome huge friction might be even better evidence that you've built something really important.
261
just used elevenlabs v4 totally on accident and i was blown away compared to v3 (just last week)
Truly blown away by the results from Eleven v4.
This model has a new architecture that unlocks more realistic and controllable speech.
You can now direct the performance of the character AND the soundscape around them. And the voice effects (like "cheap microphone") are 👌
1
26
4,784
Every major agent harness has moved 1 level up in the agentic loop flow in the last few weeks.
claude introduced projects which spawn threads
Grok bot has rooms/orchestrators which spawn an indefinite number of sub loops.
Openai is about to release o which im sure will be the same concept.
Meta just launched muse which does the same thing under the hood.
What a crazy time to be alive.
2
1
145
