lifecycle structure, discipline and constant re-validation after every step are critical when building with #ai agents.
design -> (validate) -> plan -> (re-validate) -> build (re-validate) 🔁
English
Eric Kraus
2.2K posts
Eric Kraus
@erickraus
Husband • dad • #autism supporter - Helping clients realize value with AI 🤖 X-@Microsoft • X-@Google - I chase 🌪️ in summer @mnchaser1 & big 🦈 in winter.
Minnesota انضم Aralık 2007
1.5K يتبع3.2K المتابعون
Eric Kraus أُعيد تغريده
@AnthropicAI Claude Code over there crushing it. #Agents
Lydia Hallie ✨@lydiahallie
Claude Code now supports agent teams (in research preview) Instead of a single agent working through a task sequentially, a lead agent can delegate to multiple teammates that work in parallel to research, debug, and build while coordinating with each other. Try it out today by enabling agent teams in your settings.json!
English
Eric Kraus أُعيد تغريده
Eric Kraus أُعيد تغريده
The pace of #AI tools is having a funny side effect... I've been excited to share some of the personal projects I've been working on, but I find myself re-architecting them every few days with new things! Can't even get v0.1 out!
English
The most successful organizations adopting #AI aren't the ones focused on being a first-mover...it's the ones building a foundation (and culture) for sustained iteration.
English
spent several nights last two weeks scaffolding 'Ralph' loops for dev projects. got me thinking: as #AI patterns evolve... MORE & more they seem to resemble how we 'humans' work. Is that because it's actually the best way? ...or, maybe a limitations of our own creativity/ego?
English
Had a wild 💡 ~year ago, #AI 🤖's running on simple proc. loops, looking for open tasks, async back/forth agent & user leave metadata for each other to perfect a deliverable, but NO "chat". Peers thought I was 🥜 !
Anyway... Planning Mode & Ralph Loops are super hot right now!
English
Claude Cowork - brings many of the benefits of Claude Code to the Desktop app for non-developers. It's the beginning of true 'agent-like' experiences for the average user. support.claude.com/en/articles/13… #ai
English
Having fun chatting with Sonnet 4.5 and Nano working on a new logo for my site.
English
If you don't have an assistant in 2026... You are the assistant. #ai
English
Implemented an "assignment" mechanism in my agentic framework. Now agents can create markdown "homework" for me and link to the homework file from wherever they were working. That agent will also create a task in my 'daily' note. #AI
English
There are actually many similarities between humans and AI. One of the key differences though is that humans care about our existence.
English
if you are writing #ai prompts, adding paragraphs of background 'context' to get a good response... you are doing it better than 90% of everyone else. but still doing it wrong.
English
AI agents are like top employees... capable of just about anything, but not mind readers. Top employees require clear vision, direction, context, and....continuous support. Agents are no different.
English
Eric Kraus أُعيد تغريده
Wow - ALL of Claude Code system prompts and fully up to date
This repo is an absolute goldmine, star it and fork it now
github.com/Piebald-AI/cla…
Courtesy of @PiebaldAI - the makers of tweakcc (more on that later)
English
Human thinking is naturally probabilistic - just like a lot of #AI. If you hold LLMs to human-like standards, you must give it the same opportunity (context) to be successful.
English