Sri

@imjaredz @cognition @promptlayer Congrats

Huge update.. I joined @Cognition I’m so incredibly proud of what we built at @PromptLayer. Although it feels a little surreal, I know the team is left in amazing hands. I’ll be staying on as an advisor and can’t wait to see where they take it. We started PromptLayer four years ago, and we practically coined the term “prompt engineering”. ChatGPT had come out weeks prior. Nobody knew exactly how LLMs would evolve, but we did know it will completely change the way we build. We were the first developer platform for this new type of builder. Cognition feels like an extension of this vision. Devin is how this new builder will imagine, invent, and create. We're entering the age of software abundance, to steal a term from @scottwu46 Everything is Coding Agents. It’s the single biggest problem in AI. Cognition is one of the fastest growing companies ever. But I think the team is what really sold it for me- it's packed with so many former founders Excited to share more soon...

@swyx Thank you. It's been great meeting you

@func25 I bought this earlier. Will I have access to the latest copy or do I need to purchase again?

Go is simple, so I ended up writing an 865-page book about how it works internally, just to see how it maintains that simplicity 😇

@techNmak I created a RAG using this . It works great

Someone removed the vector database from RAG and got better results. Much better. Here's what traditional RAG actually does under the hood: it chunks your document into pieces, embeds those pieces into vectors, and retrieves based on semantic similarity. The assumption is that similar text = relevant text. That assumption breaks completely for professional documents. When you ask "what were the debt trends in Q3?", vector search returns chunks that look similar to that question. But the actual answer might be buried in an appendix, referenced across three sections, in a part of the document that shares zero semantic overlap with your query. Traditional RAG never finds it. Similarity ≠ relevance. PageIndex was built around that insight. Inspired by AlphaGo, it builds a hierarchical tree index from your document - an intelligent table of contents optimized for LLM reasoning. Then it navigates that tree the way a human expert would. Not pattern matching. Reasoning. "Debt trends are usually in the financial summary or Appendix G, let's look there." What disappears: → No vector DB to build or maintain → No arbitrary chunking that breaks cross-section context → No opaque retrieval you can't explain or trace What you get: → Retrieval traceable to exact page and section references → Multi-step reasoning across document structure → Works on financial reports, legal filings, regulatory documents The benchmark: → PageIndex: 98.7% on FinanceBench → Perplexity: 45% → GPT-4o: 31% Open source.

Upcoming guests on The Pragmatic Engineer Podcast: • Thuan Pham - Uber's first (and longest-serving CTO), now CTO at Faire • Martin Kleppmann - author of Designing Data Intensive Applications • David Heinemeier Hansson (@dhh) - creator of Ruby on Rails, @basecamp and Hey • Alice Ryhl - Rust language advisor, core maintainer of Tokio (Rust's async library) and software engineer at Google • Anders Hejlsberg (@ahejlsberg) - creator of TypeScript, C#, TurboPascal • Kelsey Hightower - legendary for his Kubernetes+community work, formerly distinguished engineer at Google, minimalist Recent guests who came on the podcast: • Jean Lee (@jeanleewrites) - engineer #19 at WhatsApp, founder of Exaltitude • Steve Yegge (@Steve_Yegge) - creator of Gas Town, author of Vibe Coding, formerly at Amazon, Google • Boris Cherny (@bcherny) - creator of Claude Code, formerly one of the most productive engineers at Meta • Mitchell Hashimoto (@mitchellh) - creator of Ghostty, founder of HashiCorp • Andrey Breslav (@abreslav) - creator of Kotlin, now building the new programming language CodeSpeak • Grady Booch (@Grady_Booch) - heavily influenced object-oriented programming, creator of UML, industry legend • Peter Steinberger (@steipete) - creator of OpenClaw, previously founder of PSPDFKit • Mai-Lan Tomsen Bukovec - heads up AWS S3 • Bryan Cantrill (@bcantrill) - cofounder at @oxidecomputer, industry veteran for anything servers and hardware and software and Rust I sometimes have to pinch myself that this is real, looking through the past and upcoming guest list. Thank you to everyone listening, and to all the past and future guests for coming on the show! 🙌 Search for "The Pragmatic Engineer" on your favorite podcast player, and add it to not miss episodes. Or subscribe here: • YouTube: @pragmaticengineer?sub_confirmation=1" target="_blank" rel="nofollow noopener">youtube.com/@pragmaticengi… • Spotify: open.spotify.com/show/2Bho9xCbO… • Apple: podcasts.apple.com/us/podcast/the… • Email: newsletter.pragmaticengineer.com/about

@TheVixhal Do we have same in TS?

@dani_avila7 Best webinar today

CLAUDE.md inheritance across different sections of your project is key Great presentation from the Anthropic team at the Claude Code Advanced Patterns webinar Always good to hear directly from the people building it, how they're structuring projects and components confirms we're on the right track The direction is clear... learn to maintain all agentic components inside your repos so Claude navigates with full context awareness and controls its own context window The better you manage these patterns, the better your results

@leerob Why did you not tell before ?

Yep, Composer 2 started from an open-source base! We will do full pretraining in the future. Only ~1/4 of the compute spent on the final model came from the base, the rest is from our training. This is why evals are very different. And yes, we are following the license through our inference partner terms.

@aajay3110 @cursor_ai Is this true x.com/fynnso/status/…

I joined @cursor_ai in January to help build the future of coding agents, and I’ve been blown away by how talent-dense the team is and how much energy there is in the office. Today, we shipped Composer 2: a frontier-level model that’s fast, cost-efficient, and already part of my daily workflow. It’s also the first model where our continual pretraining produced a stronger base for the RL work that followed.

@fynnso @GergelyOrosz @swyx what's your take on this

was messing with the OpenAI base URL in Cursor and caught this accounts/anysphere/models/kimi-k2p5-rl-0317-s515-fast so composer 2 is just Kimi K2.5 with RL at least rename the model ID

You can now start a Netlify project with a prompt. Type what you want to build, pick your agent, get a live URL on production infrastructure in minutes. This is netlify.new. Here's what that actually means: ntl.fyi/46TbFce

@Netlify @jherr This is awesome

I’ve been using the composer-1.5 model since Cursor’s pricing change, and it’s been frustrating—I’m spending at least twice as much time wrestling with it, yet still not getting results that Claude-4.6-opus-high can deliver on the first try. @cursor_ai

@nicknisi This is awesome

I made a new skill: link-reader! I hate that all these good articles are written directly on twitter, which is very hostile to Claude and others actually reading it. This skill simply sees when you provide a Twitter/X link and proxies it through fxtwitter to get the actual content to Claude. github.com/nicknisi/claud…

Everyone thinks AI is a paradigm shift. That everything we've learned about building software in the last 20 years is for boomers. I disagree. That's why I built Claude Code for Real Engineers. It's a 2-week cohort that teaches AI Coding from first principles, all the way from requirements gathering to delegating to AFK agents. It's the best course I've ever built. It starts in 2 weeks, and for this week only it's 40% off. aihero.dev/s/DqHJjO

@aakashgupta Did you check pageIndex. Vectorless

If you’ve been delaying learning RAG, read this.

Everyone must learn anthropic.skilljar.com

@gregisenberg Can we not use AI Agents to help?

the truth is no matter how hard you try you’ll never be able to keep up with 100% of what’s going on in AI right now there’s just too much action right now

This is super helpful doc to know how to use claude resources.anthropic.com/hubfs/The-Comp…