Software Architect Contributes to Opensource, Eclipse JDT Committer, Strive for perfection, Neophile, Early Adapter
Joined January 2015
- Tweets 6,740
- Following 182
- Followers 188
- Likes 10,661
119 Photos and videos
Jun 11
#Anthropic always claim explicitly or implicitly their models are used by some other providers for training, they goto great extents to stop it. Still some people love their products. Twist 🧵👇
1
12
Jun 10
😂😂😂😂 non of it worked
Jun 9
fable 5 just refactored my entire codebase in one call
67 tool invocations. 1M new lines. 24 brand new files
it modularized everything. broke up monoliths. cleaned up spaghetti
none of it worked
but boy was it beautiful
12
Jun 10
Don’t hire him for a coding task. He doesn’t know to code any more. What he only knows is to write Loops 😂😂😂😂.
I pity them, trying hard to sell tokens by selling their engineering soul ❤️🩹
Claude Code’s creator
“I haven’t coded manually for 8 months. It’s been 100% Claude Code.
Right now im runing hundreds of agents with /loops and dynamic workflows.”
in just 20 minutes, Boris shows how he designs self-learning agent systems:
Loops dynamic workflows routines
That’s worth more than a $500 course on agent engineering.
1
12
Jun 10
Pretty sure none of this ridiculous features are human developed one. I’m guessing these are Bori’s Loop doings …
Jun 9
Just landed nested subagent support in Claude Code
Starting to experiment more with agents kicking off agents as a way to better manage context. Capped at depth=5 to start, going out in today’s release.
Lmk what you think!
6
Jun 9
Seems like #anthropicAI is trying to trolled on the internet with #LoopEngineering and get marketing attention. TrollMarketing ?
1
11
Jun 8
Damn these videos, make a series 📺
Jun 7
Michael Scott, the most qualified person to judge Claude vs Codex, has entered the chat.
1
6
Jun 8
These Anthropic people are on another level.
But i do see Boris is really adding loops these days, because ClaudeCode has become slow again 😂😂😂😂
42
Jun 3
In the AI era, it is frustrating see content creators acting like tech advisors just marketing on currently trending tools or topics. You never know who are they until you follow them for few weeks. #TechMadness #AIMadness
10
May 29
💯👇✅
May 28
I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem.
As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)!
I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work.
It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results?
88ms => 1.5ms
150K allocs => ~500 allocs
Incredible right? Nope.
My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path.
This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput.
The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity.
Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.
10
May 26
I want to see more like this. Currently Anthropic Executives are trying to sell them selfs in Media with false accusations of Jobs and How they run 1000s agents. Sounds like pure Marketing to me.
May 25
Microsoft is reportedly reducing internal use of Anthropic’s Claude Code after its AI bills started exploding as employee usage rapidly increased.
Some teams are now being pushed toward GitHub Copilot as the company tries to control AI costs.
Uber reportedly faced a similar problem. Executives said the company had already burned through its entire yearly AI tooling budget by April because engineers were heavily using AI coding daily.
AI coding tools are now being used for everything, and that level of usage creates massive compute and token costs when thousands of employees use these systems at the same time.
Source: TomsHardware
1
21
May 24
I might be mistaken, but I was browsing through some of the discussions on the @zeddotdev issues. I was interested in the unattended LSP binary installations and some AI Agent features. However, reading through them didn’t provide me with any clarity or reassurance. 🧵
1
22
May 24
I frequently encounter resistance without valid reasons. The author’s point is often overlooked, and the opposing arguments are presented as if they are valid, rather than acknowledging the original point. 🧵
1
4
May 24
So, I’m torn between keeping @zeddotdev editor solely because of its speed, or continuing to use it for its speed and the fact that it has a dedicated community of maintainers who actively listen to their users and provide valuable feedback. The later is missing still.
7
May 16
I wonder why #ClaudeCode doesn't allow first class support for standards like ACP so that the CLI can be used with a different UI for difference UX experience.
I don't see why #Anthropic badly wants their users to spend time in ClaudeCode and ClaudeDesktop.
29
May 16
After long time I tried Zed Agent, the user experience has improved a lot after the last time I tried it. Well done @zeddotdev team 🎉
17