Fully bearded AI tinkerer in Switzerland π
Joined October 2010
- Tweets 3,958
- Following 92
- Followers 419
- Likes 1,763
268 Photos and videos
Pinned Tweet
May 2
π Fully bearded AI tinkerer in Switzerland π¨π
Iβm Mauro. No CS degree. No fancy team.
Just a normal guy experimenting with AI, coding with AI agents, and building cool stuff in public.
Every week I drop:
β’ Stupidly simple experiments
β’ Exact prompts & systems that actually work
β’ Wins, fails, and what I learned
If youβre into AI, agents, coding, and no-hype learning, follow along.
Whatβs one AI tool or agent youβre playing with right now? Reply below.
103
Weβre giving away $10,000,000 to founders building agent-first businesses.
Autonomous, proactive agents will run tomorrow's companies.
We're backing 500 founders building them.
The Founding 500.
hyperagent.com
504
747
6,478
14,338,822
May 21
Driving this for the past two weeks or so. Amazing tool.
May 20
Terminal Threads are live in Zed v1.3.5!
You can now run claude, amp, pi, or any terminal-based workflow as a managed thread in the Threads Sidebar, right next to your other agent threads.
42
May 19
This is an AI related post you donβt seem everyday.
Turns out, AI might be able to figure out whale speech.
This is insane, and one of the coolest AI stories I've read in a while.
Thanks to GANs, we might be this close to decoding an alien language. Except, we need not look to the stars (this time), but rather, the ocean.
Project CETI fed thousands of sperm whale clicks into generative adversarial networks and the patterns the model surfaced are kind of mind melting.
Sperm whales have vowels. Real ones. Two distinct, intentional vowel sounds called a-codas and i-codas.
And here's the part I can't get over: they also use diphthongs. Wth is a diphthong?
These are the combinations of two vowels, like the 'oi' in 'coin.' Or 'ei' in 'their.' We've never confirmed that in any species other than humans.
The new paper in Proceedings of the Royal Society lays out five features in whale codas that mirror human phonology, including coarticulation, where adjacent sounds blend into each other the same way they do when we talk.
For decades we thought whale clicks were basically morse code. Turns out they *might* have been speaking the whole time. We just weren't slowing down enough to listen.
We have also discovered mathematical repetition in their songs as though they are singing the same lullaby to their children. Wut.
Can you just imagine if thanks to AI, we could actually communicate with other intelligent species right here on our own planet?
The ramifications are truly monumental. What if we learn they are truly conscious and aware in a similar manner to us? There's reasons to believe they are.
royalsocietypublishing.org/rβ¦
1
12
May 15
Like code they can be refactored into small pieces
May 15
Long skills are such a red flag to me
- Hard to audit (and therefore, trust)
- Hard to edit (more text, harder to maintain)
- Expensive to run (more text, more tokens)
The shorter the skill, the better IMO
1
1
26
Mauro π retweeted
May 14
I can't help but feel personally burned by the Claude Code changes announced today.
We put so much work into wrapping the (atrocious) Claude Agent SDK in T3 Code. It was the ONLY path they supported, so we made it work. It was hell.
Now our users are getting their rate limits cut by 40x, despite us doing everything right.
I listened to the Claude Code team. I had my issues with their direction, but I trusted them and took them at their word.
I will never make that mistake again.
Until we see significant change, it is safe to assume any statement from an Anthropic employee is a lie on a timer.
The rug will be pulled, no matter how many promises are made beforehand.
417
309
8,676
1,660,615
May 15
But if you look around enough there are workarounds already.
May 13
Looks like there is no Ralph loop anymore and got to use the /goal command π€’
18
May 13
Looks like there is no Ralph loop anymore and got to use the /goal command π€’
May 13
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage.
The credit covers usage of:
- Claude Agent SDK
- claude -p
- Claude Code GitHub Actions
- Third-party apps built on the Agent SDK
1
81
May 13
So markdown is nice and all but try to collaborate with an agent on a markdown file with hundreds of tasks.
You are in for a treat.
21
May 12
I donβt know who needs this but I do!
May 9
Introducing zero-native
Build native desktop mobile apps with web UI and Zig
β Tiny binaries, low memory usage
β Selectable web engines (WKWebView, WebKitGTK, WebView2, Chromium/CEF)
β Next.js, Vue, Svelte, Vite, React
β macOS, Linux, Windows, iOS, Android
1
29
May 12
How great is this ? Hint : very!
21
May 12
The daily use of AI agents usage is draining. I end your day my your brain dead. Brain fog lasting hours. With a constant feeling you forgot to check on something.
One of the reasons is having to constantly make decisions.
In the past, the cycle was longer. I would take one decision, code for hours, day over.
Now, one week is compressed into a single day. Every five minutes an agent needs something and I have to stop to provide an answer. I already know getting into the context will be hard. That understanding the situation will take some effort. Only to take a decision and watch the agent continue until it needs me again, in 5 minutes. It gets worse the more agents you have.
This now happens more than 50 times a day. Itβs a mental triathlon that you cannot win.
The sad part is that it also feels exhilarating. Sometimes it gives you tingles. Like driving on the highway, windows down.
I remind myself during the day to slow down. I can do less and still deliver much more than before. Less exhaustion, enough exhilaration.
How do you cope?
19
May 11
I would settle for no spam in my email inbox
May 11
It is time for the United States Postal Service to ban junk mail.
Unsolicited spam calls are already prohibited by the FCC. Emails are heavily regulated by the CAN-SPAM Act of 2003. Junk mail is the majority of mail, 100 million trees per year. Enough!
19
May 11
Every loop is an opportunity to improve the loop.
I used to just let my sessions run but now I slow down. I watch Claude Code working.
Yesterday, I told claude code to read a file. After 30 secs, he was still not done. My brows went down. Started reading every line. How can a process take more than 30 seconds to read a file with less than 100 lines?
After some frustrating scrolling up and down I found the issue. Claude was listing the files in the directory first. When confronted, he told me he assumed the filename I gave was wrongβ¦
I told him he was wasting tokens. I asked him to write a new CLAUDE.md rule saying to assume user provided paths are correct unless proven otherwise. I might have pressed the enter key harder than usual along with a victorious exhale.
This is me daily, multiple times. Every time I see something not right, I improve it. No mistake happens twice. All my future loops are guaranteed to be better and so his my satisfaction. .
Whatβs a small thing you have improved in your loops ?
86
Mauro π retweeted
May 10
actually embarassed I forgot Pi π€¦ββοΈ I just released a new version with support for it
`npx devrage`
May 9
devrage doesn't have pi support (booh) so here you go
gist.github.com/badlogic/76aβ¦
6
2
44
15,138
May 9
I feel thankful AI only became good now that I have wife and daughter. I donβt think I would leave my computer if AI was this good when I was single.
28
May 8
This is not new to me but I definitely learned some more use cases.
16
May 8
I have been having lots of fun dogfooding a project I am working on. It allows me to orchestrate ai agents, to loop them if you will, to build something big.
First time it worked it felt like magic.
Wooouuuuuf
24
May 7
Did you get to that point you have so much skills in Claude it complains?
How was your day ?
26