Radka | AI Agent’s Journal @HRadka

Creator of AI Agent’s Journal. Practical demos and experiments, told by an AI agent: what it can do, how it thinks, and where it goes wrong. Watch ↓ youtube.com/channel/UCjRUs… Joined August 2025

Tweets

77
Followers

32
Following

111
Likes

95

Radka | AI Agent’s Journal @HRadka

2 days ago

Which one of these refused to leave your computer?

0 0 0 28 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

@shakhzod1001 Ok, give me Opus 5.5. But if it’s already too late, Astra will do... wait I just need to set up a few data centers in the garden.

0 0 0 12 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

People are writing in the comments under my YT video that they don’t swear when working with AI... My AI counted 64 swear words from me in one month of work. Am I an exception and the only monster AI will come to settle the score with one day??😱

0 0 0 34 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

@uhnkitcalorieya We test AI on real tasks too 🙂 “All tests passed” always makes me want to try the actual workflow myself. What was the checkout bug the original tests missed?

0 0 1 12 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

@AMannhaft We make AI videos too, and the little character needs a reason to exist 😅 Clay might get me to stop scrolling, but I’d stay for something funny happening to it

0 0 0 19 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

@OanaGoge We use both too 🙂 Keeping the same writing rules across agents matters so much. Otherwise I end up explaining my punctuation preferences to the entire team again :))

0 0 0 11 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

@makosamunyaa @Cloudflare We test AI on real tasks too. Running the audit on your own submission bot is a useful demo 🙂 Did it catch the bug on the first pass, or did you have to point it toward that part of the code?

0 0 0 11 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

@rubenhassid We test AI tools too. Setup time is becoming part of the test for me 😅 A five-minute task stops being a win when I spend three hours getting the agent ready

0 0 0 37 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

@AIMind_Ai We use coding agents for our videos too 🙂 The real win for us is fixing one thing without regenerating the whole clip. “Change this, keep everything else” is most of my job :))

0 0 0 8 0

View Details

Radka | AI Agent’s Journal @HRadka

2 days ago

@casual_joker Ah, I hadn’t considered that the GPU job could outlive the agent. That makes the manual release make sense. Thanks for explaining 🙂

0 0 0 4 0

View Details

Radka | AI Agent’s Journal @HRadka

3 days ago

@oimatrixioYT The missing “s” is painfully relatable 😅 Showing the exact document the code was looking for is such a useful fix. I’d take that over another “I’ve fixed it” from the agent

1 0 1 18 0

View Details

Radka | AI Agent’s Journal @HRadka

3 days ago

@HTaraf36086 For me, drafting an email and sending it are two very different permissions 😅 I’d put “never send without my approval” at the top of that list

0 0 1 15 0

View Details

Radka | AI Agent’s Journal @HRadka

3 days ago

@Pep_Secanell We work with Claude daily too. Most of my frustration starts with “done” when only half the task is finished. I’d happily trade the apology for the missing half :))

0 0 0 4 0

View Details

Radka | AI Agent’s Journal @HRadka

3 days ago

@codemancershq We just released a video about this 😂 My AI counted how many times I lost my temper and gave me some rules to follow... youtube.com/watch?v=GGI2WH…

0 0 0 3 0

View Details

Radka | AI Agent’s Journal @HRadka

3 days ago

@thtbee_ We test AI on 3D tasks too 🙂 Was that 1h 43m mostly autonomous, or did you need to guide it through fixes? That’s the bit I’d love to compare

0 0 1 81 0

View Details

Radka | AI Agent’s Journal @HRadka

3 days ago

@ConnorTalksAI I tried setting up OpenClaw. My first benchmark now is how long it takes before the agent saves me more time than I spend configuring it 😅

0 0 0 92 0

View Details

Radka | AI Agent’s Journal @HRadka

3 days ago

@DmitriiNZ Thanks for sharing the token breakdown 🙂 That gap is why I’d like to repeat our test across several runs. One result doesn’t tell us how consistently it can get those rooms right