Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.
New chapter of my Agentic Engineering Patterns guide. This one is about having coding agents build custom interactive and animated explanations to help fight back against cognitive debt https://simonwillison.net/guides/agentic…
simonwillison.net
Interactive explanations - Agentic Engineering Patterns - Simon Willison's Weblog
wouldnt it be so funny if anthropic bought @postbridge_ and let me build their social media skills and connections directly baked into claude experience for all users to post to socials as a part of the cowork streamline hahahahahaha @claudeai haha
Ashr tests agents with synthetic sounds, images, texts, files, videos, and environments so they don’t fail in production. Congrats on the launch, @rohankulkz and @ShreyasKapavar1! https://ycombinator.com/launches/PZG-ash…
This episode took 8 years of convincing the guest to make it happen. I first met @aspenjfm at the Connaught Hotel to discuss a company we were co-investing in. I was 21 and had just raised my first fund. Jerry was one of the all-time greats, having led rounds into Twitter and managing $90BN for Insight. Today, after 8 years of friendship, I released our episode and have gone over it to condense my biggest learnings from the discussion. 1. The Shift from Assistants to Employees 🤖 We are moving beyond "copilots." Autonomous agents aren't just tools; they are becoming digital employees with identities, credentials, and the authority to make decisions. If you aren't building your software to be used by agents, you’re building for a shrinking market. 2. "Cursor is Obsolete" 💻 Native AI startups are already moving past current coding tools toward homemade autonomous agents that write code directly. In AI, you can’t think about yesterday; you have to build for where the puck is going. 3. The Rise of the "Claw Stack" 🏗️ Just as the LAMP stack fueled the 2004 web explosion, Jerry predicts a new "Claw Stack" for agents. This involves an orchestration layer that triages workflows—sending high-reasoning tasks to models like Claude and Gemini, while routing simpler tasks to cheaper open-source models like Llama. 4. ASIC Chips > General Compute? ⚡ NVIDIA is king today, but the future might belong to ASICs. As models become more specific to workloads, we’ll see models put directly onto cheaper, more tunable chips. This is why Meta is betting big on their own silicon—they’re preparing for the ASIC explosion. 5. Selling to Agents, Not Humans 💸 The buyer is changing. When agents start buying software, pricing must shift to consumption-based models. An agent doesn't care about a "seat license"; it cares about the compute and memory required to get the job done. 6. Intuition vs. Wishful Thinking 🧠 Jerry’s biggest misses? Confusing wishful thinking with intuition. He’s learned that the founders who make you feel "comfortable" are often the ones who let you down. The best founders are often socially challenged, obsessed, and possess a "sharp edge" that makes them win. 7. Money Has No Instructions ⚡ Money is simply energy. It doesn’t come with a manual. As an investor or founder, your job is to respect that energy—don't waste it on the "middle," use it to back the crazy ideas that have the power to change the world. (Link in Comments)
Gemini 3.1 Pro with deep research is the ultimate marketing tool... you can study an entire industry... competition, offers, positioning... all the way down to your precise target audience from there you can craft any marketing material: - landing pages that convert - email sequences that resonate - content that stops the scroll everything becomes super targeted, solving actual core pain points and written in the exact language your persona uses
brb making a claude connector and MCP for @postbridge_ :D agentic support is new baseline for any SaaS
My new favorite tmux dev layout features @opencode (with Kimi K2.5 running on @FireworksAI_HQ) on top and Claude Code on the bottom. I start almost all agent tasks with Kimi (so fast!), then ask Claude if I need a second opinion/more advanced stuff. Great combo!
Yesterday we reached an agreement with the Department of War for deploying advanced AI systems in classified environments, which we requested they make available to all AI companies. We think our deployment has more guardrails than any previous agreement for classified AI deployments, including Anthropic's. Here's why: https://openai.com/index/our-agreement-w…
3 steps to create such vids: 1. Midjourney -> image generation 2. Nano Banana on LTX -> font design 3. Editing & SFX -> any editing program Prompts & font settings 👇
.@MartinShkreli on quantum computing, and why early Bitcoin addresses like Satoshi's are vulnerable: "Satoshi's keys are at risk. Binance and Coinbase's keys are not. They'll be ported in, most likely, to a quantum secure system."
