Writing about AI & SaaS Founder, Serpyn - Collaboration software for SaaS
- Tweets 9,559
- Following 798
- Followers 3,365
- Likes 11,174
Pinned Tweet
I’ll keep this post PINNED until I hit 10K followers. 🔥🔥
> Indie SaaS builder & Dad of 2 kids (I love them so much ❤️)
> Passionate about AI and writing about it. I’m constantly building and shipping AI SaaS products, as well as productivity and business mobile apps.
>I want to make more friends. The solo builder journey can get pretty lonely, so come hang out and let’s support and motivate each other! 🤝
4
18
1,096
Hey @sama, when is OpenAI becoming OpenSI?
(Before the IPO, I assume? Gotta upgrade the name before upgrading the valuation. 😂)
1
4
111
If AI companies can legally train models on copyrighted internet content and books, why was The Pirate Bay hunted worldwide for facilitating P2P sharing of copyrighted material? ☹️
2
48
The next PC OS may not be Windows vs. Linux. 🤔
> It may be an OS with two layers: one optimized for humans, one for AI agents.
1. Humans get the GUI, apps and controls they expect. 2. Agents get a sandbox with direct access to files, processes, APIs, terminals and compute.
(Same machine, two interfaces)
> The future OS may not ask: “Who owns the computer?”
It may ask: “Who is using it right now?”
2
56
Don’t get fooled by this bar chart. 😅
> Almost 98% of US households still aren’t paying for AI. It’s wild that the world’s AI superpower still has a population this early in the adoption curve. If the US is only at 2.2%, imagine the rest of the world.
> The opportunity is still insane. Every builder on X is gonna be a millionaire. 😂
98% of US households aren't paying for AI yet
More charts in State of Markets II: a16z.news/p/state-of-markets…
3
2
129
Okay, now even North Korea is putting AI into its military arsenal. 😬
> What does an AI-powered missile actually do? Smart trajectory changes to make interception harder? China is rapidly pushing AI and robots into its military. What about the US? 😅
> The more I watch this unfold, the more those sci-fi movies I grew up with start looking less like fiction and more like product demos.
48
Apparently, AI hardware is now valuable enough to get you arrested. 😂
> A California computer-company owner was arrested for allegedly smuggling $300M+ worth of export-controlled AI servers to China.
> AI is now a national weapon, and hardware is its ammunition. In the US–China AI race, cutting off compute is a brutal move: without hardware, you can’t keep training models or scale them to more users.
3
77
Builders, be honest: if over the next year you still don’t make a single dollar from your work (or you make less than you spend on AI), would you still keep paying for it? 😅
3
5
127
This is surprisingly close to Elon Musk’s vision of an abundance economy. 🙂
> More compute → cheaper AI and robotics → lower production costs → cheaper essentials.
> My bet is that humans will gradually shift from production toward services, creativity, and experiences. > Wealth itself may be redefined, not by how much money you earn, but by how much of life’s necessities you no longer need to worry about paying for.
Anthropic's Sholto Douglas says AI capex could rise from $1 trillion this year to $4 trillion in 2028 and begin doubling humanity’s GDP by the early 2030s
“Over the course of the last four or five years, we’ve been 2x-ing or 3x-ing the amount of compute capacity devoted to AI every year."
"And so I think that’s roughly, together, the hyperscalers are spending about a trillion dollars this year on capex related to AI. A very interesting question will be, will that trend line hold, and so will it go to $2 trillion next year and then $4 trillion in 2028?"
"And if that trend line broadly holds, and you can expand that to encompass the broader robotics industry, then that means that in the early 2030s, you start to get to the point where you’re actually effectively doubling the GDP of humanity, which, again, is a little bit of a ridiculous concept, but requires a lot of things to go right.”
5
133
Sam Altman warning people not to worship AI is peak 2026 comedy. 😂
> You can’t spend years selling “superintelligence that may outperform humanity,” then act surprised when some users start treating the chatbot like the Oracle of Delphi, turns out hype has side effects too.
I am very uncomfortable about people trying to ascribe religious force or a surrender of human judgment to AI models, and think it is a real safety issue.
1
2
308
The chart below shows a typical website’s traffic-source mix today.
> How do you think this pie chart will look in 2027? (What will gain, what will shrink, and why? 🤔)
2
5
139
Building a SaaS today:
"100% Supabase" ...or.. "PostgreSQL + VPS"? 🤔
> Which one would YOU choose, and why?
I’m especially interested in real-world cases, not textbook answers.
5
3
119
Supabase isn’t buying another database, it’s preparing for a world where AI agents create millions of apps, each needing isolated data. 🤔
> Turso makes disposable databases cheap enough to scale, while Postgres handles production, giving Supabase a path from an agent’s first experiment to a real business.
2
3
102
For those using OpenAI’s Dots or Grok bots:
> How do you prevent your entire agent squad from going completely off-scope while executing your original task? 🤔
57
Snapchat disappeared like it never existed. Did anyone actually use it? 😂
(Maybe Snapchat should’ve taken Facebook’s acquisition offer after all. 🤣)
1
2
153
Well, that makes perfect sense!!! Who better to advise Anthropic ahead of a historic IPO than someone who once pitched Jeffrey Epstein to fund an adult-content startup? 😂
> At this point, people are naturally going to wonder what other high-profile connections Cami Clark may have, but that’s a very different claim from having evidence of them.
🚨𝗥𝗨𝗠𝗢𝗥: Anthropic CEO Dario Amodei’s wife Cami Clark (who infamously pitched Jeffrey Epstein to fund her "luxury porn" startup) has stepped down as key advisor to the CEO, weeks before IPO
1
4
169
Trump may have accidentally discovered a new way to shill domain names: forget crypto, now he’s pumping Slovenia’s internet. 🤣
> After he rebranded AI as “Super Intelligence,” .si registrations exploded: 11,000 in a single day,
> while domain squatters immediately started listing “superintelligence” .coms for $20K;
> somewhere in Slovenia, a domain reseller is probably already calling this quantitative easing.
Have you registered any .si domains yet? 😅
1
200
OpenAI doesn’t need another feature, it needs a better architect for the whole AI experience ☹️
(because Dot feels like the right idea wrapped in a product nobody quite knows who it’s for).
> Grokbot still carries Elon’s fingerprints: simple, practical, direct, and genuinely ahead on cloud agentic workflows;
> but I still believe OpenAI will iterate Dot into something much stronger, because with models as capable as Astra/Sol, fixing the architecture and orchestration could unlock a huge jump almost overnight.
quick comparison between OpenAI’s "Dot" and Grokbot: I prefer Grokbot.
With Grokbot, I have a Chief of Staff, a daily team meeting where all my Grok bots exchange information, and various sessions dedicated to different tasks. This not only provides a better overview but also helps synchronize the various activities.
OpenAI’s Dot might well be very potent - especially given the capabilities offered by Astra - but I haven't quite grasped exactly who it is initially aimed at. Is it for the casual user or for more professional users? The design feels more playful- geared more toward ChatGPT users than Codex users. At the same time, however, OpenAI is heavily promoting the power of its "always-on" agents.
Once OpenAI’s Dot is further developed and allows for delegating tasks to multiple agents, I’ll give it another try.
1
99
I made a stupid mistake, and Apple rejected my app because of it. 😭
> I intentionally created a fake sensitive scenario to demonstrate that my app could refuse s**ual-relevant requests,
> but the test content remained visible in the UI and the reviewer might reject it for containing sexual content. 😅
> Learn from my mistake: "never leave your safety-test content exposed in the review build".
1
1
77