Michal Stanislawek
@xmstan.bsky.social
Strategist & Solution Builder
Conversational & Generative AI
Live Media
Conversational & Generative AI
Live Media
Your prompts only make sense in your head. Here's a test that'll prove it.
My prompts are full of hidden assumptions. Yours are too. "Respond naturally" doesn't really mean anything without context.
🧵 Time for some prompt therapy.
My prompts are full of hidden assumptions. Yours are too. "Respond naturally" doesn't really mean anything without context.
🧵 Time for some prompt therapy.
June 23, 2025 at 5:48 AM
Your prompts only make sense in your head. Here's a test that'll prove it.
My prompts are full of hidden assumptions. Yours are too. "Respond naturally" doesn't really mean anything without context.
🧵 Time for some prompt therapy.
My prompts are full of hidden assumptions. Yours are too. "Respond naturally" doesn't really mean anything without context.
🧵 Time for some prompt therapy.
The fastest way to burn your AI budget? Chasing the 'autonomous agent' hype.
Here's the saner, more pragmatic path 🧵
Here's the saner, more pragmatic path 🧵
June 21, 2025 at 6:30 AM
The fastest way to burn your AI budget? Chasing the 'autonomous agent' hype.
Here's the saner, more pragmatic path 🧵
Here's the saner, more pragmatic path 🧵
I accidentally broke Gemini 2.5 Pro today. And instead of crying, it *thought* of its own bug report. Well, sort of...
Here is what happened 👇
Here is what happened 👇
June 20, 2025 at 1:27 PM
I accidentally broke Gemini 2.5 Pro today. And instead of crying, it *thought* of its own bug report. Well, sort of...
Here is what happened 👇
Here is what happened 👇
Playing buzzword bingo with another "AI Expert"? Use these 5 questions to flip the table.
"AI Experts" are everywhere. But talk is cheap. To find real builders - the ones who've actually shipped a project and have the scars to prove it - you need to ask questions.
🧵👇
"AI Experts" are everywhere. But talk is cheap. To find real builders - the ones who've actually shipped a project and have the scars to prove it - you need to ask questions.
🧵👇
June 20, 2025 at 5:57 AM
Playing buzzword bingo with another "AI Expert"? Use these 5 questions to flip the table.
"AI Experts" are everywhere. But talk is cheap. To find real builders - the ones who've actually shipped a project and have the scars to prove it - you need to ask questions.
🧵👇
"AI Experts" are everywhere. But talk is cheap. To find real builders - the ones who've actually shipped a project and have the scars to prove it - you need to ask questions.
🧵👇
Using MCP Servers? Be careful out there...
👇 Here are the 5 MCP security issues everyone should be aware of
👇 Here are the 5 MCP security issues everyone should be aware of
April 6, 2025 at 4:43 PM
Using MCP Servers? Be careful out there...
👇 Here are the 5 MCP security issues everyone should be aware of
👇 Here are the 5 MCP security issues everyone should be aware of
Is MCP becoming AI's Super Glue?
Imagine staring at Blender's interface. Those countless buttons, menus, and options that leave most people overwhelmed with creative anxiety...
🧵 A thread 👇
Imagine staring at Blender's interface. Those countless buttons, menus, and options that leave most people overwhelmed with creative anxiety...
🧵 A thread 👇
March 27, 2025 at 5:33 PM
Is MCP becoming AI's Super Glue?
Imagine staring at Blender's interface. Those countless buttons, menus, and options that leave most people overwhelmed with creative anxiety...
🧵 A thread 👇
Imagine staring at Blender's interface. Those countless buttons, menus, and options that leave most people overwhelmed with creative anxiety...
🧵 A thread 👇
Everyone talks about AI agents. But what is the real spectrum of agency? 🤔
@CobusGreylingZA invited me to dive deep into the realities of implementing AI. We'll explore all from the evolution of convo design in the age of AI agents to the nitty-gritty of running LLMs in prod.
@CobusGreylingZA invited me to dive deep into the realities of implementing AI. We'll explore all from the evolution of convo design in the age of AI agents to the nitty-gritty of running LLMs in prod.
February 17, 2025 at 3:12 PM
Everyone talks about AI agents. But what is the real spectrum of agency? 🤔
@CobusGreylingZA invited me to dive deep into the realities of implementing AI. We'll explore all from the evolution of convo design in the age of AI agents to the nitty-gritty of running LLMs in prod.
@CobusGreylingZA invited me to dive deep into the realities of implementing AI. We'll explore all from the evolution of convo design in the age of AI agents to the nitty-gritty of running LLMs in prod.
A new chapter in the cloud wars? That's the $500B question...
The Stargate Project ($500B investment over 4 years) isn't just another tech headline - it might be OpenAI quietly positioning itself as the next cloud giant.
Let me break this down for you... 🤔
The Stargate Project ($500B investment over 4 years) isn't just another tech headline - it might be OpenAI quietly positioning itself as the next cloud giant.
Let me break this down for you... 🤔
January 22, 2025 at 11:21 AM
A new chapter in the cloud wars? That's the $500B question...
The Stargate Project ($500B investment over 4 years) isn't just another tech headline - it might be OpenAI quietly positioning itself as the next cloud giant.
Let me break this down for you... 🤔
The Stargate Project ($500B investment over 4 years) isn't just another tech headline - it might be OpenAI quietly positioning itself as the next cloud giant.
Let me break this down for you... 🤔
Wieczorne Śniadanie AI na strzelnicy?! Tak, to nie pomyłka – technologia przeplatała się tu z hukiem wystrzałów, tworząc niezapomniane wrażenia.
W poniedziałek odbyło się specjalne, wieczorne lubelskie Śniadanie AI na strzelnicy.
W poniedziałek odbyło się specjalne, wieczorne lubelskie Śniadanie AI na strzelnicy.
December 12, 2024 at 1:00 PM
Wieczorne Śniadanie AI na strzelnicy?! Tak, to nie pomyłka – technologia przeplatała się tu z hukiem wystrzałów, tworząc niezapomniane wrażenia.
W poniedziałek odbyło się specjalne, wieczorne lubelskie Śniadanie AI na strzelnicy.
W poniedziałek odbyło się specjalne, wieczorne lubelskie Śniadanie AI na strzelnicy.
Does +$180/mo premium makes any sense? 🧐
It might... Here's why 👇
o1 preview was already pretty great at coding
o1 will surely continue that.
o1 pro mode will do fractionally better at math or other tasks.
Do I care? Nah!
But...
It might... Here's why 👇
o1 preview was already pretty great at coding
o1 will surely continue that.
o1 pro mode will do fractionally better at math or other tasks.
Do I care? Nah!
But...
December 5, 2024 at 11:49 PM
Does +$180/mo premium makes any sense? 🧐
It might... Here's why 👇
o1 preview was already pretty great at coding
o1 will surely continue that.
o1 pro mode will do fractionally better at math or other tasks.
Do I care? Nah!
But...
It might... Here's why 👇
o1 preview was already pretty great at coding
o1 will surely continue that.
o1 pro mode will do fractionally better at math or other tasks.
Do I care? Nah!
But...
Czy na strzelnicy da się wdrażać innowacje i wykorzystywać AI?
Zdecydowanie tak - i to w jakim wydaniu!
Ale zamiast opowiadać i teoretyzować, zawsze lepiej spróbować.
Zdecydowanie tak - i to w jakim wydaniu!
Ale zamiast opowiadać i teoretyzować, zawsze lepiej spróbować.
December 5, 2024 at 7:50 AM
Czy na strzelnicy da się wdrażać innowacje i wykorzystywać AI?
Zdecydowanie tak - i to w jakim wydaniu!
Ale zamiast opowiadać i teoretyzować, zawsze lepiej spróbować.
Zdecydowanie tak - i to w jakim wydaniu!
Ale zamiast opowiadać i teoretyzować, zawsze lepiej spróbować.
The most effective way to stop AI scraping your content?
David Mayer
😆
David Mayer
😆
December 2, 2024 at 6:03 PM
The most effective way to stop AI scraping your content?
David Mayer
😆
David Mayer
😆
Jailbreaking language models isn't just fun and games anymore... it becomes a ticking time-bomb.
Convincing LLM powered robots to ignore safety rules and potentially cause harm to humans is far too easy.
Convincing LLM powered robots to ignore safety rules and potentially cause harm to humans is far too easy.
November 25, 2024 at 10:17 PM
Jailbreaking language models isn't just fun and games anymore... it becomes a ticking time-bomb.
Convincing LLM powered robots to ignore safety rules and potentially cause harm to humans is far too easy.
Convincing LLM powered robots to ignore safety rules and potentially cause harm to humans is far too easy.