Moritz Kremb
@moritzkremb
I tinker with AI and share what I learn along the way | German, Engineer, Ex-PM, creator | Free resources→
709 Following 75.8K Followers
agencies are just as screwed as software
look at what’s happening to video agencies right now. so many creators i know have either fired their video agency or just not hired one if they were planning to
claude can do it all for you now. better, faster, cheaper.
and i think that’s going to happen to most other agencies too
why pay for an agency to run your meta ads if AI can do it better? why pay for someone to manage your influencer campaigns if grok bot can do it?
all the value will accrue to the AI labs in the short-term until new opportunities are created in the market
Show more
I'll be setting up my Notion brain today
Trying to build the perfect system that works locally, in the cloud and across harnesses
while keeping everything in sync
New: Give all your agents one shared brain in Notion.
Now, ChatGPT, Claude, Muse, and Bot can save what they learn to the same memory and use the same skills in Notion.
Learn how to Notion-pill your agents:
Show more
this makes so much sense for notion to do, and they’re in the perfect position to do this
New: Give all your agents one shared brain in Notion.
Now, ChatGPT, Claude, Muse, and Bot can save what they learn to the same memory and use the same skills in Notion.
Learn how to Notion-pill your agents:
Show more
AI can now improve itself
Frontier LLMs were found to be able to read and write neural network weights! recursive self-improvement could now be close.
> what this means?
> why it matters?
Full breakdown and explanation for normies:
Show more
I built framejam for video editing. but i just realized it's super useful for editing html presentations as well
you can just add your feedback on your slides and then let your agent make all the changes at once
Show more
X now started suggesting reels in my feed
They’re currently also running a creator program where you get paid $5k to post 15 videos in 30 days
videos are going to get pushed hard on this platform
Show more
this is huge
researchers found that the latest AI models can surgically edit another model's weights to improve it. without re-training it.
if this scales, things are about to go crazy
made a short video to explain
Show more
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog:
Show more
in 2023, GPT-4 could work on its own for only about 4 minutes
today the best models handle tasks that would take a human weeks
METR has been tracking this and the curve is wild:
> from 2019 to 2025, the length of tasks AI agents could finish doubled roughly every 7 months
> since 2024 it's been closer to every 3 months
we take it for granted now, but here's what changed just in the last year:
> models got way better at reasoning
> they got better at using tools
> they catch and fix their own mistakes instead of spiraling
> they test their own output repeatedly until it's correct
strap in. we're only just getting started
Show more
Unpopular opinion: AI basically became useful for enterprise use early this year with agentic AI acceleration. Everything before was basically consumer and hobbyist. 2026 is year 0 for AI enterprise use. So why would anyone have expected measurable productivity increase already.
Show more
grok bot now has a cap on how many skills you can save 🤨
ChatGPT in Germany in 2027
this is like putting ads on facebook before it was cool
they will stifle the industry before it even has a chance to take off
JUST IN: German minister proposes a new tax on all economic value created by AI.
elon just said longevity is coming.
i've been digging deeper lately into how to prepare for it.
it all sounds kinda crazy now, but i think all of this will be pretty normalized in the next couple of years with how fast AI is evolving.
dying in the next 10 years or so, just before immortality becomes possible, would be the most tragic timing ever imo.
so i want to do everything i can to give myself and my loved ones the best chance of making it there.
here's how i'm preparing.
the must-haves for me:
1. cryonics
you pay a company to preserve your body or brain after death until technology exists to revive you.
in the US, there are several providers. the biggest one is Alcor. in Germany, the relevant provider for me is Tomorrow Bio.
i haven't set this up yet, but i'm getting everything ready now.
i always assumed you needed a huge amount of cash upfront. but it's actually not that bad. here's the breakdown:
Tomorrow Bio costs €50/month for membership and €200k for whole-body preservation (it's similar for Alcor).
the membership basically makes sure a team is on standby to respond when you die and prepare you for transport and preservation. (that's just 2 netflix subscriptions. worth it if you think about it.)
the €200k sounds like a lot at first. but that can actually come from life insurance.
if you're in your 30s like me, term life insurance could cost around €10–20/month. if you die before the policy ends, say at 65, it covers the €200k.
meanwhile, you would build up €200k on the side. if you outlive the policy, you can pay the fee yourself.
for my parents, who are past 65, i'd focus on having the funds ready directly. at that age, you could also get whole life insurance, but it would cost several hundred per month.
for my kid, it's trickier. some providers don't accept under-18s. Alcor and Cryonics Institute in the US do, though, so i'll probably set it up with one of them until he's 18 and then see.
another option is a much cheaper provider called Cryonics Institute. they charge only around $28k. but they basically only do the freezing, and you have to organize the transport yourself. and the long-term funding is less secure, which means they might run out of money at some point, and your relatives would then need to pay to keep you frozen. although this might not even be such a big issue at the speed at which everything is evolving now.
2. cord blood + tissue banking
i think this is a total no-brainer if you're planning to have a kid.
we did this for our son through Vita 34 in Germany, and i'm super glad we did. and we'd of course do it again if we have more kids.
after birth, they collect blood from the umbilical cord and placenta, plus a piece of cord tissue. then they freeze it to save the stem cells.
the idea is to keep those cells available for medical uses, including potential future therapies. so it's kind of an insurance even before the cryonics insurance.
for us, it cost €1,670 upfront + €180/year for storage. so totally doable.
in the US, there are many providers too, like CBR, ViaCord and Cryo-Cell.
some other things i'm looking into:
→ sperm / egg freezing: keeping the option of having more kids later.
→ iPSCs: reprogramming blood or skin cells into stem cells for potential future repair. still mostly research, but interesting because my own cord blood wasn't stored.
→ blood banking: freezing a sample for potential future testing or making stem cells.
→ microbiome banking: preserving your gut bacteria for possible use after heavy antibiotics.
→ baby-tooth / dental stem cells: banking cells from inside baby teeth or wisdom teeth.
again, i think it all kinda sounds crazy now.
but with how fast AI is evolving, i think the next few years will make it much clearer how possible this future is becoming.
and i want to give myself and my family the best chance of being around for it. i think it's actually kinda stupid not to.
Show more
@FischerKing64 Longevity is coming
met up with @JGrandillon today!
we talked about the future of content, AI and what’s ahead
thankful that we both decided, around 3 years ago, to just pick up our phones and post about the things we find interesting
Show more
try
@moritzkremb Loving this tool! Use it everyday, & I have shared it with others
2 feature requests:
1. Have a descript-like side-by-side with text and video rather than having text just in the timeline. Want to highlight segments of the video by highlighting text.
Show more
fantastic explanation of post-labour economics
Okay little Uzi, I've been working on this for a while, so lemme break it down for you.
The primary thing you're worried about is what's called "household income" - which is how families get money to buy groceries and pay bills. 🥑
Now, jobs and work are the PRIMARY way that households get money right now. But not the ONLY way!
There are actually THREE sources of household income: jobs, government checks, and property. 💹
Like, how some people get welfare and retirement, that's government source.
And some people get money from renting out houses they own and dividends from stocks. That's property.
Now, as you rightly figured out, jobs are going away. In fact, jobs have been losing power for decades. But the core problem is household income.
For the last few centuries, the primary way has been the wage/labor distribution mechanism. In other words, the primary way people participate in the economy is to sell their labor for wages. 🏧
It's been this way for SO LONG that people can't really imagine another way.
But there is!
It's called stock/flow, which is a fancy way of saying "live off rent and dividends" which is actually how most rich people live today.
I call this Post-Labor Economics. Why? Because labor is going away!
Now, you might wonder "well how do we get started if we're not rich already?" (This is the most common question!)
The answer is that it takes some time, and there are actually many ways that the system, government, and companies can all help create this new economy!
One example is to create what's called a Sovereign Wealth Fund. This is like a gigantic stock portfolio that the government runs on behalf of all the people, and pays them dividends, usually on an annual basis, or quarterly. That means you get some money just from being a citizen! 🇺🇸
But labor isn't gone yet, so there are other options to get more capital. One is for employees to own the companies they work for. In America, this is called an ESOP, and I'm England an EOT, and there are plenty of other countries with similar programs like cooperatives.
Now, those are just two kinds of programs, but there are many, many more like them all over the world. Citizen dividends, cooperatives, trusts, rebates, and more!
Eventually, what we can build is a layered portfolio, so that every household gets checks every month, quarter, or year, from many sources. National, local, and more. AND, on top of that, old school property can stick around, like rent and dividends from common stock.
As you correctly point out, without consumer demand, the economy sort of implodes 😅
So, it's in everyone's interests to figure out Post-Labor Economics!
There are a lot of other changes that need to happen for this to work, but just because there are many moving parts, doesn't mean it's impossible! For instance, right now, payroll tax and income tax provides most of the funding for the government. Well, if labor and wages go away, so does most of the government funding! Oops!
That means we would need to reconfigure the tax code to take more from companies and those who profit from AI and robotics and the automation that takes the jobs!
Now, some people worry that we simply won't have enough resources to go around, that this can't possibly work. But think about it this way: if AI and robots and automation are good enough to take all the jobs, then they are probably good enough to make all the food, houses, and other stuff we need!
Hope this helps 😁
Show more
the ai bio wars are heating up
this will be the next big thing in ai
what's been happening in case you missed it:
> oct 2025 - anthropic launches claude for life sciences
> apr 2026 - anthropic buys coefficient bio (reportedly ~$400m, <10 ex-genentech scientists)
> apr 2026 - openai launches gpt-rosalind, it's bio model
> may 2026 - isomorphic labs (alphabet's ai drug co, deepmind spinout) raises $2.1b and eyes first human trials
> jun 2026 - nobel laureate john jumper (alphafold) leaves deepmind for anthropic
> sep 2026 - anthropic confirms its own wet lab in the bay area
> oct 20 - openai acquires orchestra bio (ai agents for trials, timelines, cros) and folds it into rosalind
> oct 27 - anthropic reveals acquisition of mithril: a 1-person biotech run on claude. starts in-house drug programs. human trials possible by late 2028
frontier labs are becoming biotech labs. extremely exciting times ahead
Show more
Over the past few months, Anthropic scientists have engaged in deep research partnerships with six organizations developing new medicines. We've also begun drug discovery efforts of our own.
In doing so, we’ve learned much more about how Claude can support the many stages of drug development.
In this blog post, we share some early results from these efforts, and describe how this work fits in with our wider ambitions in the life sciences.
Show more
New Grok Bot Template
I made a Grok Bot that lets ChatGPT, Claude Code or any other agent send tasks to your Grok Bot.
How it works:
> add the bot, it gives you a webhook URL, key and one ready-made prompt
> paste that prompt into ChatGPT or Claude Code
> it then sets up the /send-to-grok-bot skill
Get it here:
Show more
Met the legend @kimmonismus tonight!
Grok @bot has gotten so good lately
It's a huge difference from just a few weeks ago
In this video:
>All the latest Grok Bot updates
>My top use cases and how my set up looks
>Can it run autonomous businesses now?
Full breakdown:
Show more
Lots of new faces in the last few days!
So let me reintroduce myself.
> I'm Moritz
> Half-German, Half-Taiwanese
> Grew up in Beijing and Singapore
> I've always loved software so I became a product manager
> Became a dad 3 years ago
> Decided that working for someone isn't my thing
> Started creating content about AI
> Turned it into a business
> Founded several software startups along the way. Many failed. Some worked.
Now I spend my days trying out the latest AI tools, creating content about it, building products, and helping businesses implement AI.
My latest obsession is creating AI videos. So figured, why not turn this post into a video... here you go 😄
Show more
