Discover the Best AI Tools

We test the top AI tools for writing, video, images & music — so you don't have to.

Explore AI Tools

OpenAI's Decisions API Is Now Live for Everyone — and It Was Built to Catch a Startup

Image
  Most AI answers arrive as essays. OpenAI's new Decisions API doesn't write a word — it answers yes or no, picks from your categories, or scores on your rubric, in 100 to 180 milliseconds. It went generally available this week, and the backstory is the best part: OpenAI's own product lead, Nikunj Handa, reportedly admitted the product wasn't on the roadmap four weeks ago. A small startup forced the sprint. Yes or no, in 100 milliseconds The API runs on GPT-6 Luna. Developers send text or images plus typed questions and get machine-readable answers in three shapes: a probability that a condition is true, a pick from a fixed list with a probability for each, or a score on rubric levels you define. Input costs $0.10 per million tokens, and output tokens are free. It launched as a public beta on October 6 (after a limited DevDay preview in late September), and the general release adds voice and visual input plus geographically targeted inference. Early customer Speak, the...

Anthropic's Claude Duet Is an AI That Messages You First

Anthropic Claude Duet proactive AI agent cover art



Anthropic logo
Anthropic just flipped the script on how AI assistants work. Instead of waiting for you to type a prompt, its new app Claude Duet messages you first — with a morning briefing, nudges when something needs your attention, and background work that continues even after you close the app.

The details
Announced October 21, 2026, Claude Duet is a research preview rolling out gradually to Claude Pro ($20/month) and Max (from $100/month) subscribers in the US, through a waitlist. It ships as standalone iOS and Android apps and on the web at claude.ai — at no extra charge, though Anthropic warns background activity counts against your plan's existing usage limits and could burn through them "sooner than with chat alone."

Duet runs on Claude Opus 5.5, picked by Anthropic specifically for its judgment about what deserves your attention and when to ask for approval. It shares regular Claude's memories, skills, and connectors, can search your past conversations, and operates on its own cloud-based computer — so tasks keep running while your phone is off. Its capabilities include a Tasks tab for ongoing assignments, Stripe integration to flag unpaid invoices and draft reminders, and Link/Shop Pay for purchases with your approval. And Anthropic is explicit on the guardrails: Duet is designed to ask before it sends a message, makes a payment, signs a document, or deletes anything.

The honest fine print
Team and Enterprise plans are excluded from the preview (expansion is promised later), the rollout is US-only for now, and Anthropic itself cautions that Duet can misread a thread, misjudge urgency, and can't yet show you its browser. Check-ins happen every few hours rather than continuously — deliberately, to avoid draining usage limits faster.

Why it matters
This is the third shot fired in the proactive-assistant war, arriving weeks after OpenAI's Dotsand while Meta's Muse keeps expanding. The technology is converging; the winner will be whoever gets the judgment right — interrupting at exactly the right moments without annoying you into muting it.

The take
An always-on assistant that respects your attention is genuinely exciting — and genuinely hard to build. Duet's cautious, approval-first design is Anthropic's classic play: slow, safe, deliberate. Whether that's fast enough to matter against Meta's and OpenAI's broader launches is the real question.

Comments

Popular posts from this blog

AI Tools Every Accountant Should Use in 2026

Top 5 Best AI Video Generation Tools in 2026

7 Best AI Website Builders in 2026