running https://xitter.cf/t.co/vG9keEzuol and constantly rebuilding it around what AI makes possible. Writing about AI, operations, people and what actually works.
Warszawa, Polska
Joined June 2015
- Tweets292
- Following2.1K
- Followers254
- Likes13K
Marcin Warno retweeted
Let’s be clear about what’s happening here: Ukraine is being threatened with starlink removal and intel being cut off for literally fighting back and defending itself. For nearly
5 years the west has been trying to get Ukraine to hold back. Utterly pathetic.
Marcin Warno retweeted
If confirmed, this is telling: Russian data centres get hit, and online activity around AfD suddenly plummets. 📉
So much talk about German patriotism… then the connection from Russia goes down. 🤔🇷🇺
Marcin Warno retweeted
Here's a 45-second TL;DR on Jev.
I find the core idea beautifully simple, but the video made it really hard to understand.
Hope you find it helpful.
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
Just watched the @webflow Keynote and my first reaction was: fuck yeah, they’ve started building exactly what we described in our manifesto 😆
They announced "Source", an agentic development platform that’s supposed to combine company context, marketing tools, and building and managing websites (something simmilar what you’d get from ploy/codex/claude (with the right setup).
The surprising part is that the code is meant to be the source of truth, and the Source is supposed to work not only with Webflow build but also custom code (Astro, Next etc.)
Sounds cool, BUT:
1. We got almost no specifics other than sexy launch video
2. The product is only in limited research preview, and nobody knows when it will become widely available
3. Nobody knows what the pricing will look like
4. Webflow says the OG Webflow isn’t going anywhere (we all know how that sounds considering how E-commerce, Logic, Memberships ended up), but it’s unclear how the current Webflow and Source will ultimately coexist.
Based on what they showed, it’s easy to start wondering why anyone would build a new website in the current Webflow today if their new platform and vision is also supposed to work with existing custom-coded websites haha.
What do you think?
The keynote: youtube.com/watch?v=B8BFooHf… (watch from 1:06, the first part is mostly marketing BS)
Marcin Warno retweeted
So, my thoughts on the Webflow Conf keynote.
I think it's an understatement to say Webflow fumbled this whole AI transformation.
But fumbling doesn't mean the play is over.
For the past few months, it's felt like the ball was going to hit the ground at any second - but, some of their recent blog posts (ie: 'getting to the source') showed me that the ball wasn't quite on the ground yet.
This conference, to me, felt like the moment where Webflow was either going to catch the ball, or we would all watch it hit the ground.
So... the keynote. To begin, we got 40 minutes of new and updated features - although, notably, they were mostly features targeting marketers in enterprise teams. Fine. I was there in the chat telling everyone to let them cook.
Then... the last 20 minutes - introducing, Source by Webflow.
Your own codebase. Your own stack. Your own CMS. All in one desktop app. Sounds like something I'd like, perhaps even something I'd build 😉
So - they caught the ball?
Honestly, if Webflow released this back in January, we might be looking at a very different reality. But they didn't. In fact, they didn't even release it today - it's in 'research preview', and will be 'rolled out to select customers over the next few months'. Ouch.
See... 'a few months' in this market is an eternity. Within 'a few months', Ship Studio was built from scratch, tested, launched, and put on the machines of hundreds of agencies and freelancers.
What seems cool today will not seem cool in ~3 months. Remember app gen? AI assistant?
However, even if it was launched today... there's something HUGE missing.
Pricing.
Let me give you an example. Ship Studio is cool. Thousands of people use it every day. But if I put a paywall on it, even $20 a month, 99.9% of users would leave.
Why? Well... because you can get the exact same thing without Ship Studio. Frontier coding agents like Codex, tried tested and true enterprise governance with GitHub, AI-accessible product analytics with PostHog, enterprise-grade hosting with Vercel... you get the idea.
The moat is gone - the workflow outlined by Webflow in their demo of Source is truly the future, I agree - but this isn't a new thing.
Now, Webflow has announced their existing product to be the old way, and their new alternative isn't available for use yet... and when it is, the competition won't be Framer - it will be free and open source tools... and believe me, you can't compete with free.
Anyways - the point of this post isn't to say that Webflow's ball hit the ground. It's an honest post on my reaction from me, someone who's used the product for 8 years, and knows and loves dozens of people who rely on Webflow's continued success.
And to Webflow's team - I am more than happy to share any and all feedback with you amazing people. This isn't over just yet. But I cannot stress enough to you how important it is that Source ships quickly.
Julian out 🫡
Marcin Warno retweeted
I will close one of these tabs today and never look back
@webflow 🤔 @astrodotbuild
Marcin Warno retweeted
There is no reason NOT to have every single email you get routed through an agent / LLM / API to be scored and approved or rejected
And then the approved emails ordered by importance
It's crazy this isn't a feature of Gmail, Fastmail etc
AI could permanently fix email!
Marcin Warno retweeted
Copy-paste components that you can directly use in Claude / Codex!
beautifului.dev by @theShaneLevine
aicss.dev by @kvnkld
transitions.dev by @Jakubantalik
amicro.vercel.app by @SubhanHQ
canvasui.dev by @davidhaz
What would you add?
Marcin Warno retweeted
- GrokBot UI
- GPT Sol level model
- With an in app browser
Would be insane.
Grokbot (Huge fan) is held back by Grok 4.6.
I fall for AI hype trains all the time 🤠. Most disappoint me within a few hours.
Grok @bot didn't.
@spacexai entered the personal agent race and got the UX right on the first try.
Here's why:
1. The context lives with the agent
With ChatGPT Work, Codex and Claude Cowork, I still have to think about the setup before I can actually use the agent.
Chat? Project? Local? Cloud?
With Grok Bot, I create the agent and use it.
One day in, I already have separate agents for personal brand, delivery, our website, finance and paperwork, plus a Chief of Staff.
I just open the right one and continue.
2. Connecting tools is stupidly simple
You send it a link. It gives you an auth link. You log in. Done.
3. It's actually persistent
I can close my laptop, continue from my phone, and the same agent is still there with the same context.
If it needs me, I take over for a minute and hand it back.
It's not perfect. The model itself feels weaker than what I'm used to in Claude or Codex, and some basic workflow stuff is still missing. You can't even queue another message while the agent is working.
Most of what Grok Bot can do was already possible somewhere else.
I just don't have to figure out how to make it all work anymore.
If there's a personal agent that does this better, tell me (@viktor_com doesn't count. I love it, but that's a different category)
Marcin Warno retweeted
IT IS TIME FOR EUROPE AND URSULA WAN WEIDEN TO RISE ON TOP WITH LE CHATON FAT*
Marcin Warno retweeted
Brand + web sprint for revenuebase.ai/
40 components, 10+ pages, dark & light mode
delivered in 6 weeks
