Dima Vasilenko

I build agentic systems - MCP servers, autonomous workflows and tooling that runs unattended.

[Dima Vasilenko @malzernBO]

@savipww Stop the cap

[Dima Vasilenko @malzernBO]

@Bober_smart Stop the cap

[Dima Vasilenko @malzernBO]

Ledsa er ute i App Store! 🥳

Dette er en milepæl jeg har gledet meg skikkelig til.

Ledsa startet fordi jeg selv så hvor unødvendig vanskelig det kan være å finne ut hvor ledsagerbevis gjelder, hva ledsager betaler og hvilke vilkår som gjelder.

Det ble starten på https://t.co/0yLvRldUsi.

Nå er Ledsa også blitt en egen app for ledsagerbevis på iOS.

I appen kan du søke etter steder i hele Norge som godtar ledsagerbevis, utforske steder i kartet og sjekke pris, vilkår, kilde og når informasjonen sist ble kontrollert.

Målet er fortsatt det samme: Informasjon om ledsagerbevis skal være enkel å finne når du faktisk trenger den.

Ledsa er gratis. Ingen konto. Ingen annonser.

Og vi er bare i gang.

Ledsa kan lastes ned i App Store nå:

https://t.co/DMrPsLl2tG

#Ledsa #Ledsagerbevis #Tilgjengelighet #AppStore #iOS

[Dima Vasilenko @malzernBO]

@0xboan Stop the cap

[Dima Vasilenko @malzernBO]

@0xCristal Stop the cap

[Dima Vasilenko @malzernBO]

@0xDezo Stop the cap

[Dima Vasilenko @malzernBO]

@rimtoln Stop the cap

[Dima Vasilenko @malzernBO]

@Dan1ro0 Stop the cap

[Dima Vasilenko @malzernBO]

@0xObssnnn Stop the cap

[Dima Vasilenko @malzernBO]

@RoundtableSpace Stop the cap

[Dima Vasilenko @malzernBO]

@bl888m_eth Stop the cap

[Dima Vasilenko @malzernBO]

Ledsa kommer til App Store! 🥳

Apple har nå godkjent Ledsa for distribusjon, og appen blir tilgjengelig for iOS på fredag.

Søk etter steder som godtar ledsagerbevis, se hva ledsager betaler og når informasjonen sist ble kontrollert.

Gratis. Ingen annonser. Ingen innlogging.

https://t.co/0yLvRldUsi

#ledsagerbevis #iOS

[Dima Vasilenko @malzernBO]

My workload is unusually compute-intensive: autonomous agents, continuous research, validation, orchestration, APIs and MCPs. Grok Bot is extremely capable, but sustained high-frequency workflows expose the economics pretty quickly. I’m optimizing the architecture to separate routine execution from premium inference.

[Dima Vasilenko @malzernBO]

@WaitWhat @bot Im paying 300$ a month for supergrok heavy and reached max in 24 hours. Its not worth it as it is today.

[Dima Vasilenko @malzernBO]

@0xrux Atleast try to give something that actually works for the community :p

[Fekri @fekdaoui]

not hating btw, grok @bot is the best agent i've used so far

but it's not really usable for most ppl right now, the rate limits are just way too low

on top of that, there's like 8 different subs and it's not clear at all how much more usage each one actually gives you

[Dima Vasilenko @malzernBO]

@peteystruggles I completely understand what your saying! Im going to test something new tonight and i will post about it if its a success!✌️

[Dima Vasilenko @malzernBO]

I’ve used OpenClaw, Codex, Claude and Grok Bot and more.

Grok Bot is nice but the pricing makes no sense for heavy usage

I’m paying $300/month and burned through my credits in 12 hours

Now I have to wait 7 days for a reset.

Hard to call agents 24/7 when usage lasts half a day

[Dima Vasilenko @malzernBO]

@_duyet Its not gone but i would not recommend. After 12 hours with work between 5 agents credits went to 100% usage and there is a 7 day reset timer.

[Dima Vasilenko @malzernBO]

I’ve spent the last few days building an autonomous AI-powered crypto trading system from scratch.

This isn’t a ChatGPT wrapper that outputs BUY/SELL.

I wanted to see how far I could push an agentic architecture where AI operates as part of an actual trading system rather than being the trading system itself.

The stack now includes:

- Multiple Grok agents with separated responsibilities
- Automated market research and signal analysis
- Deterministic execution engine
- Risk management and position controls
- Strategy versioning
- Backtesting infrastructure
- Paper/shadow trading against live market data
- Performance attribution and strategy metrics
- Persistent trade and decision history
- Monitoring + failure handling
- A custom dashboard for observing the entire system in real time

The important architectural decision was separating AI reasoning from execution.

Grok can research markets, analyze signals, challenge strategies and generate recommendations but it doesn’t get unrestricted control over capital.

Every proposed action still has to pass deterministic risk and execution rules.

I’m also running strategies through backtests and shadow execution before allowing them anywhere near real capital.

The goal isn’t to build an AI that predicts every move.

It’s to build a system that can continuously:

observe → research → hypothesize → test → execute → measure → improve

without requiring me to sit in front of charts all day.

The interesting part starts now.

Can an autonomous multi-agent system actually develop a measurable edge after fees, slippage and bad market regimes?

I’m going to document the results publicly including the failures.

No fake PnL screenshots. No cherry-picked trades. No “AI money printer.”

Just the architecture, methodology and results.

#alexfinn #AI