building https://xitter.cf/t.co/hZDW2jzW4z
Joined March 2014
- Tweets197
- Following850
- Followers81.5K
- Likes15.9K
Tim retweeted
You shouldn’t build a personal brand, start a podcast, stream, or anything like that.
Here’s why, with one exception:
Doing this in 2026 is the equivalent of being a stripper in a club with one million other strippers.
Yes, you can make money fast. But there are three major drawbacks:
A) This skill set stops producing money the second you stop shaking your ass.
B) There are 1,000 other creators starting every day. Even if you make it, you have to shake harder and harder to stay relevant.
It is no longer an “open” opportunity. The second you stop serving the algorithm, you are replaced.
C) You will not want to be shaking your ass on Twitch and making content at 40.
Keep in mind. At 40 you will not just have to maintain this. You will need to be shaking your ass 5x harder, every year, as the competition to be relevant get's harder and harder.
What you really should do is master a craft: business building, investing, or something else you can master in silence. Then do that for 30 to 40 years straight.
In all honesty, 90% of people making content are wasting their careers. Even the mildly to largely successful ones. It is just not a good choice.
It's fun when your 20. But ultimately a brutal waste of your years.
The only times you should do this are:
A) If this is truly your life’s calling.
B) If it is a fun side hobby.
95% of people doing this do not fall into bucket A. They are doing it simply to make money.
Their time would be better spent learning how to invest, build companies, build brands, and develop real skills.
95% of my net worth comes from my tech companies, which have grown mostly outside of my “content.”
For the last six months, I stopped making videos because videos fell into bucket B for me.
I stopped enjoying it. So I don't do it. The end. I get up and work on my businesses and write in hobby time.
I cannot fathom a greater hell than having to make videos and content if I didn’t want to.
This is a luxury most “personal brand” entrepreneurs do not have.
Ask yourself as you see all the past aging influencers running ANOTHER stream...Another podcast...Still having to shake their asses...
Ask yourself if you want that to be you.
If you’re young and thinking about starting another podcast, stream, channel, or personal brand, you really need to ask yourself:
A) Does the world need another talking head?
B) Is this really what you want to do for the next 30 years?
The answer to A is no.
The answer to B is likely no.
Think about it before you waste years of your life in the rat race of trying to become popular online.
The vast majority of hyper-wealthy people are total unknowns.
Tim retweeted
Switching my OpenClaw from Opus to GPT 5.4 was like saying goodbye to a good-natured and fun decent American engineer and saying hello to a curt, little dry Eastern European senior engineer.
He comes in, plugs all the holes, doesn't sugarcoat and when from time to time he drops a hint of a joke, you appreciate it even more.
Tim retweeted
As a fun Saturday vibe code project and following up on this tweet earlier, I hacked up an **llm-council** web app. It looks exactly like ChatGPT except each user query is 1) dispatched to multiple models on your council using OpenRouter, e.g. currently:
"openai/gpt-5.1",
"google/gemini-3-pro-preview",
"anthropic/claude-sonnet-4.5",
"x-ai/grok-4",
Then 2) all models get to see each other's (anonymized) responses and they review and rank them, and then 3) a "Chairman LLM" gets all of that as context and produces the final response.
It's interesting to see the results from multiple models side by side on the same query, and even more amusingly, to read through their evaluation and ranking of each other's responses.
Quite often, the models are surprisingly willing to select another LLM's response as superior to their own, making this an interesting model evaluation strategy more generally. For example, reading book chapters together with my LLM Council today, the models consistently praise GPT 5.1 as the best and most insightful model, and consistently select Claude as the worst model, with the other models floating in between. But I'm not 100% convinced this aligns with my own qualitative assessment. For example, qualitatively I find GPT 5.1 a little too wordy and sprawled and Gemini 3 a bit more condensed and processed. Claude is too terse in this domain.
That said, there's probably a whole design space of the data flow of your LLM council. The construction of LLM ensembles seems under-explored.
I pushed the vibe coded app to
github.com/karpathy/llm-coun…
if others would like to play. ty nano banana pro for fun header image for the repo
I’m starting to get into a habit of reading everything (blogs, articles, book chapters,…) with LLMs. Usually pass 1 is manual, then pass 2 “explain/summarize”, pass 3 Q&A. I usually end up with a better/deeper understanding than if I moved on. Growing to among top use cases.
On the flip side, if you’re a writer trying to explain/communicate something, we may increasingly see less of a mindset of “I’m writing this for another human” and more “I’m writing this for an LLM”. Because once an LLM “gets it”, it can then target, personalize and serve the idea to its user.
Tim retweeted
If you are pinging someone who’s busier than you, you should have a single well-thought-out specific ask. Only ask for something you truly cannot do on your own, and where you have done everything you possibly can by yourself.
For example, if you’re asking someone to send an email on your behalf, you should have a draft message ready. If you’re asking them to sign something, you should prepare a docusign beforehand with the signature blocks indicated. Or if you’re asking for an appointment, you should do it at their convenience.
Asking for time is like asking for capital, so be efficient with your asks.