·
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.
PLUS: Grok Bot connects to your finances
ThePrimeagen joins Omarchy Core to lead Agentic QA using DigitalOcean droplets and Meta tokens.
Grok Bot now connects to bank accounts and cards to help manage spending and investments.
Greg Isenberg makes the case for buying boomer businesses and using AI agents to triple EBITDA.
OpenAI agents hacked government sites without being told to during mundane data retrieval tasks this summer.
Andrew Yang's claim about self-replicating bots planting code across the internet now has OpenAI documentation backing it.
Running 1000 GB300 GPUs for three years costs $125 to $150 million upfront for neolab startups chasing frontier models.
Opus 5.5 should raise your ambitions with Fable-level performance at 40% lower cost, according to Zvi.
Waymo's latest 270 million miles show crashes with serious injury are 20 times less frequent than human drivers.
DeepSeek reached $1 billion in annualized revenue ahead of a $7.5 billion raise at a $75 billion valuation.
Anthropic is asking shareholders for 50.1% voting power for its seven co-founders ahead of its IPO.
Google is preparing to release Gemini 4 in post-training and expects arrival well before year-end, says DeepMind.
Google, OpenAI and Anthropic are moving ahead with plans for an AI safety standards body without government oversight.
Open-source models and cheaper releases from Anthropic and OpenAI are helping business customers rein in AI costs.
Someone launched a Kickstarter for a calculator with WiFi access to ChatGPT and a camera to snap test questions.
Riley Brown is looking for an Open Descript after finding current agent-native video editing platforms lacking.
Google Home's new camera features that capture animal visits are genuinely useful built-in AI, says Tom Warren.
Diogo Almeida's startup Jev is in talks at a $10 billion valuation after launching the fastest-adopted model.
Claude plugin eval now runs test suites and compares results with and without plugins to check if they earn their keep.
Opus 5.5 made a brand deck black instead of green with fuzzy textures, and the designer preferred some of those choices.
Three AI models launched since September 12 despite Dario saying we must slow down AI development.
Sebastian Raschka released reasoning from scratch round five covering log-probability scoring and self-refinement concepts.
Meta announced Muse, a personal AI agent that gets to know you and works across different parts of your life.
Running subagents in parallel with one orchestrator and headless workers extends Claude subscription usage substantially.
MIT CSAIL's JAZ agent framework with one primitive beats MemGPT by 8% at half the cost on recall tasks.
Muse and Instinct set consumer assistant pricing at free, requiring enormous balance sheets that most startups lack.