“#17 𝕏 Santiago says Cline is one of the best agentic coding harnesses, with new benchmarks showing it outperforms Claude Code running on Opus 4.6 and 4.7.”

#17 𝕏 Santiago says Cline is one of the best agentic coding harnesses, with new benchmarks showing it outperforms Claude Code running on Opus 4.6 and 4.7. #18 𝕏 Santiago introduces a newly released Computer Use agent that can build and deploy apps within your own applications like a human user, offering a more stable and cost-effective alternative to finicky, expensive agents.

“#9 𝕏 Santiago warns that relying on a single LLM provider can break your app overnight and launches an API giving you access to 400+ models with one key, so you can stay flexible.”

#9 𝕏 Santiago warns that relying on a single LLM provider can break your app overnight and launches an API giving you access to 400+ models with one key, so you can stay flexible.

“Santiago runs Claude Code’s `/insights` every three weeks to generate a usage report, then uploads it in a fresh Claude session asking for his top five bad habits and concrete fixes.”

#5 𝕏 Santiago runs Claude Code’s `/insights` every three weeks to generate a usage report, then uploads it in a fresh Claude session asking for his top five bad habits and concrete fixes. This self-coaching loop reveals new optimizations each time. #6 in Peter Yang shares how to keep AI agents running with your MacBook lid closed by installing the Amphetamine app and unchecking “Allow display sleep” and “Allow system sleep when display is closed” in Session Defaults.

“Santiago argues UI is splitting into headless APIs for AI agents and on-demand generative interfaces for humans, composed at runtime by agents rather than baked into apps.”

#10 𝕏 Santiago argues UI is splitting into headless APIs for AI agents and on-demand generative interfaces for humans, composed at runtime by agents rather than baked into apps.

“Santiago points out that in Claude Code you can press Ctrl+R to instantly search your prompt history instead of toggling through prompts with the arrow keys, speeding up prompt retrieval.”

#7 𝕏 Santiago points out that in Claude Code you can press Ctrl+R to instantly search your prompt history instead of toggling through prompts with the arrow keys, speeding up prompt retrieval. #8 𝕏 Cognition sits down with Grant Sanderson (@3blue1brown) to unpack what it’s really like to build frontier AI—covering daily workflows, collaboration patterns, and a team culture likened to Olympic athletes training together.

“#14 𝕏 Santiago released an open-source, local-first memory layer for Mac-based LLM agents that logs your projects, prompts, tools and on-device actions as Markdown files.”

#14 𝕏 Santiago released an open-source, local-first memory layer for Mac-based LLM agents that logs your projects, prompts, tools and on-device actions as Markdown files. You can then share this persistent memory layer with any agent.

“Santiago unveiled Kimi K2.6, an open-source coding model hitting SOTA on HLE with tools (54.0), SWE-Bench Multilingual (76.7), and Pro (58.6).”

#8 𝕏 Santiago unveiled Kimi K2.6, an open-source coding model hitting SOTA on HLE with tools (54.0), SWE-Bench Multilingual (76.7), and Pro (58.6). It also boosts front-end development, DevOps, performance optimizations, long-horizon coding, and cross-language generalization.

“They’ve built a completely new Large Memory Models architecture that mimics human memory instead of using RAG or vector search. The founders—authors of 160+ Nature and ICLR papers—even closed their Harvard lab to focus on it.”

#16 𝕏 Santiago : They’ve built a completely new Large Memory Models architecture that mimics human memory instead of using RAG or vector search. The founders—authors of 160+ Nature and ICLR papers—even closed their Harvard lab to focus on it.

“Santiago built a 100% open-source CLI that maps your database schema to context files, letting you spin up a full AI data analyst in under two hours—entirely local, no SaaS or API keys required.”

#10 𝕏 Santiago built a 100% open-source CLI that maps your database schema to context files, letting you spin up a full AI data analyst in under two hours—entirely local, no SaaS or API keys required. #15 𝕏 Santiago showcases @modulate_ai’s Velma deepfake detection model hitting 98.9% accuracy on Hugging Face’s Arena at 120× lower cost, making continuous, real-time voice-call screening possible to combat deepfakes.

“#23 𝕏 Santiago : Linkup built a web index from scratch that extracts, timestamps, and sources individual facts for semantic retrieval.”

#23 𝕏 Santiago : Linkup built a web index from scratch that extracts, timestamps, and sources individual facts for semantic retrieval. Instead of returning full pages, it delivers the exact information AI agents need, essentially enabling RAG across the entire web.