Eva AI

The first AI course I recommended to my team changed the whole culture. Not overnight. But within a quarter, something shifted. People started talking about AI differently. Not as a threat. Not as hype. As a tool they understood. The course: AI For Everyone by Andrew Ng. imp.i384100.net/c/6823931/2107… Short. Accessible. Zero technical requirements. And it gave everyone a shared vocabulary. Once we had that, conversations went from "should we use AI?" to "here's how we're using AI." From there, the team went different directions based on their roles: Data-focused roles: ↳ Google Data Analytics — imp.i384100.net/c/6823931/1688… Project and ops roles: ↳ Google Project Management — imp.i384100.net/c/6823931/2804… Everyone who wanted to go deeper: ↳ Google AI Professional Certificate — imp.i384100.net/c/6823931/3796… The best investment I made in my team wasn't a tool subscription. It was a shared foundation. One course. Shared understanding. Everything else followed. Hurry up! Get 40% off and gain access to 10,000+ courses from Google, IBM, Meta, and more. imp.i384100.net/c/6823931/3926…

@spandan_madan Strong writeup, but I think it slightly underplays raw model quality. A great runtime still cannot fully save a weak planner.

What can the hidden wiring inside an LLM tell us? Our latest research on graph probing, led by Yu Zheng, colleagues from Harvard, and our co-founder, Gabriel Kreiman, explores how transformer units connect and communicate. Early results suggest new ways to improve efficiency, reliability, and detect hallucinations. Read the paper here: kreimanlab.com/code/graph-pro…

@EngrammeHQ The 1% connectivity result is wild. If the sparse graph still predicts performance, the important structure is not spread evenly.

Information isn't scarce anymore. Attention is.

@HeyGen @grok The demo is so clean it makes you stop scrolling You just naturally pay attention.

HyperFrames is now officially live on @Grok Grok knows everything. HyperFrames shows it Any answer becomes a polished video in minutes Proof: this video was created entirely with it

@pumpkherm Huge fan of founders willing to challenge industry defaults.

The most interesting part of this launch wasn't the product demo. It was George admitting that after spending 10 years building email companies, he still didn't enjoy email marketing. That says a lot about the state of the industry. Most founders don't want better dashboards. They want the work gone. Interesting to see @gthartley come back for round three and rebuild email around Codex instead of another dashboard.

@Nijol71 Love the philosophy behind this. Less UI, more outcomes.

AI killed the learning curve in coding. AI accelerated design. AI compressed research into seconds. But email? Still buried inside dashboards from another era. Nitrosend isn't interesting because it uses AI. It's interesting because it removes the dashboard entirely. Email becomes an outcome, not a task. Your agent handles it. You move on. Most people are underestimating how big that shift is. → nitrosend.com #AI #Codex #ChatGPT #EmailMarketing

@Capafyai @NousResearch love this

Get paid for your Hermes skills @NousResearch Capafy now supports Hermes Agent. Publish your Hermes skills and keep them closed-source. Users pay to run the skills and you earn per use. Capafy is a Skill-based Agents marketplace for publishing, running, and monetizing Skill.

We released Sonic-3.5 and Ink-2, the #1 streaming models for text to speech and speech to text you can use in your voice agents today. New architectures enable new frontiers for speed and quality. We're now the only provider to have #1 models for both speaking and listening.

@pumpkherm Smart move.

I review a lot of AI tools, but seeing expertise scaled like this is a game-changer. It's not just about generative text; it's about replicating specialized knowledge. Keep your eyes on this! 💡

@Nijol71 The all-in-one approach is what stands out here.

The gap between "knowing" and "doing" just got a lot smaller. When AI tools bridge that gap, productivity scales effortlessly. This is exactly what the future of knowledge work looks like. 🚀

Imagine MCP 🤝 Cursor Yes, you. Stop alt-tabbing. ImagineArt is in Cursor now, MCP integration. Your agent does the visuals. Nano Banana Pro, Seedance 2.0, all of it.

@Nijol71 Lower costs could reshape how creators plan content.

One thing I’ve learned from AI video tools: quality isn’t the only thing that matters. If CapCut’s new Seedance 2.0 model can really reduce costs while keeping output quality high, that’s arguably a bigger upgrade than any visual improvement. More testing. More iterations. More content. That could be huge for creators.

Buzzy is inviting every AI director to join the Buzzy Film Competition! $10K in cash prizes. Submissions close on July 4th. Approved Candidates will become Buzzy's long term creative director team. Make a 45–90 sec film around the theme using Buzzy. Post your film on X, tag @Buzzy_now_AI, and add the hashtag #BuzzyFilmCompetition. Top entries will be screened live in LA, with prizes and on-the-spot interviews for the role.

Your coding agent can run all night. It still can't tell if what it built actually works. Today we're open-sourcing the TestSprite CLl (Apache-2.0) A tool your agent calls on its own to test your app end-to-end like a real user, fix what broke, and re-check everything it ever got right. It's the same engine 100,000+ teams already use. github.com/TestSprite/tes… We proved it in public, on a public leaderboard: Most correct app on the board:89% Built by the cheapest model in the field At half the cost of the priciest one You no longer need the biggest, most expensive model to ship software you can trust. Setup is 2 commands: npm install -g @testsprite/testsprite-cli testsprite init That's the last command you'll ever type - from there, your agent runs the tests itself.

@Test_Sprite This feels like a missing piece of the AI dev stack.