Follow AI Builders — Not Influencers

"AI won't take your job, but someone using AI will"

RT Marmotize Looks like the current best performing combo for agents is: - file system - code execution - skills interfacing with cli-s/MCPs That is why I am currently bullish on deepagents from LangChain. @hwchase17 @huntlovell Is there any guide out there for using deepagents with AWS Bedrock AgentCore Runtime? Original tweet: https://x.com/marmotize_/status/2020229358309503212

Parsing line charts is a hard task for VLMs VLMs are generally fine at coarse visual understanding, but they have a hard time reasoning about precise coordinates. Ask most VLMs, even though tuned to chart understanding, to parse a line chart to a table and they will struggle. We tested over a few samples. Docling’s new granite-vision model, gemini 3 flash, gpt 5.2 pro, and a v0.1 of our own chart parsing (which is in beta and rapidly evolving). Out of these, most models fail, and sometimes miss the entire chart correctly. gpt 5.2 pro is closest but spends an absurd number of tokens reasoning through each point. Our own parsing is actually quite good, though of course, there’s still some things we need to do to get to 100% accuracy. If you want to parse complex documents with diagrams/charts, come check out LlamaCloud! https://cloud.llamaindex.ai/

1/8 for anyone curious about the nuts & bolts (maybe to build their own), here's a quick comparison of openclaw, babyagi3, and nanobot https://github.com/openclaw/openclaw https://github.com/yoheinakajima/babyagi3 https://github.com/HKUDS/nanobot (one-shotted by claude: https://claude.ai/public/artifacts/9687c464-32d0-4697-b7bf-a9595c8b7403) 01/ at a glance2/8 02/ core architecture3/8 04/ tool system4/8 03/ memory systems5/8 05/ communication channels6/8 06/ deployment & operations7/8 07/ codebase quality8/8 08/ summary comparison https://x.com/yoheinakajima/status/2020027037180932347?s=20

Big game, Nano Banana world 🏈 Open the Gemini app and type: “Miniature world scene featuring an extreme close up of tiny football players [insert action] on a football field, stadium background, High-quality macro photography. AR 1:1” Drop yours in the replies. ↓

I bet no one with kids under 6 fall into this category 😅

RT Chubby♨️ 1/ Thankfully, I was able to test Perplexity's new "Model Council" and I was honestly curious. I put it through a rigorous test, which aims to examine the socio-economic consequences of AGI on Western society (see prompt in the comments) using empirical data and its own conclusions. What I really like: Each model writes its own longer report, which can be read individually. Afterwards, a table compares which statements are agreed upon and where differing opinions prevail (screenshot below). In addition, each model's unique discoveries are listed. Seriously: a really cool feature for conducting research without having to run each model individually and then compare them. I'll continue testing it. But so far, I really like it! (no paid collaboration. I did not receive any money for testing.) Original tweet: https://x.com/kimmonismus/status/2020213054412321052

RT LangChain Claude Code 🤝 LangSmith Curious what Claude Code is doing behind the scenes? Or want observability into critical workflows that you’ve set up with Claude Code. With our new Claude Code → LangSmith integration, you can view every 🤖 LLM call and 🔧 tool call Claude Code makes. Docs: https://docs.langchain.com/langsmith/trace-claude-code Video: https://www.youtube.com/watch?v=NoSmSVKMSMs Original tweet: https://x.com/LangChain/status/2020209201708969986

RT Claude Our teams have been building with a 2.5x-faster version of Claude Opus 4.6. We’re now making it available as an early experiment via Claude Code and our API. Original tweet: https://x.com/claudeai/status/2020207322124132504

RT Harrison Chase Traces are cool Original tweet: https://x.com/hwchase17/status/2020198906211209532