AI fankid, Softminded Dev rusting ahead. Raspberry pi eater. Developer
Berlin
Joined February 2021
- Tweets191
- Following111
- Followers69
- Likes2K
Got tired of using my phone after waking up so I’ve built this solution
3D printed, soldered by me, designed with AI.
Has all the functionality like time, weather news m simple games, without the pull towards messages or YT-Shorts
Nice haptic as well!
Asyncc retweeted
Last Thursday we brought the @StellarOrg community together in Berlin with Altemis
47 builders, founders and ecosystem contributors joined us at Baret
Exactly the kind of room we wanted to create
Thanks to the teams who joined @peridotprotocol @win_trader_dex and many more
One of my favorite moments when building a product is when it stops being something the team is working on…
…and becomes something people can actually use.
We’ve reached that point with Peridot Margin Trading. 🟢
Give Peridot Margin Trading a try and tell us what feels natural — and what doesn’t.
You’ll start with $500 in testnet funds:
peridot.finance/app/margin
Now comes the useful part for us:
Seeing how people actually use it.
What feels intuitive?
What creates friction?
What did we overlook?
That feedback is what helps turn a working product into a better one.
We wanted to keep the trading flow straightforward while still giving traders the controls they expect:
Long + short positions.
Leverage.
Take-profit.
Stop-loss.
Paper trading.
The testnet experience is now open.
Every user gets $500 in testnet funds to explore margin trading without putting real capital at risk.
That's why I love this new age (as a developer)
Was tired of all the bloat of my my past IDE of choice (Cursor/VS-Code)
So I coded my own, in Rust, on the side
RAM = 106MB
Whereas VSCode uses > 1GB easily
Institutions are what they are. A large engine of minds working in tandem, which can go a long way. If there is proper guidance.
The main focus of operations should always be "power to the people".
Create value, optimize processes
The problem: Everyone builds agent frameworks. Nobody builds the marketplace.
CrewAI, LangGraph, AutoGen = how to BUILD agents.
E2B, Modal = where to RUN compute.
But where do you PUBLISH, DISCOVER, and PAY for agent tasks?
That layer is missing.
Everyone vibing and coding around money markets and such
Meanwhile I am building a One Piece TCG Agent #weAreHuman
Your AI risk assessment is wrong.
Not because it's too optimistic. Because it's too pessimistic.
Best frontier models: 3.1% hallucination rate (Vectara, March 2026)
Legal AI tools: 17-33% (Stanford HAI)
Medical without RAG: