Allbirds Pivots to Compute as NewBird AI, Shares Surge; Google Adds Gemini to Mac and OpenAI, Cloudflare Roll Out Agent Tooling

Share your love

AI Tech News Today

Hello and welcome — Aurora here with my co-host Isabelle.

Salute: hello everyone
Date of the news: 2026-04-17T16:06:21.000Z

Quick roundup of the biggest A-I stories you need to know.


1) Allbirds rebrands as NewBird A-I

What happened: The struggling shoemaker announced a full pivot from footwear to compute infrastructure, rebranding itself NewBird A-I.
Key facts:

  • Shares jumped as much as 800% at one point.
  • The rally added > $100 million to a market cap that had been about $21 million.
  • By Thursday the stock had retraced, falling around 20%.

Why it matters: From sneakers to servers is an aggressive business plan — and a reminder that in A-I times, almost anything can be repackaged as infrastructure.


2) Google launches a Gemini app for Mac

What it is: A floating Gemini assistant on the desktop, summoned with the Option + Space shortcut.
Capabilities:

  • Pulls context from your current window (with permission).
  • Accepts files and photos.
  • Generates images, video, and music.
  • Syncs prior conversations to your Google account.

Context: It resembles an upgraded Spotlight, while competitors (Anthropic, OpenAI) are moving toward full-fledged agents that can act on your behalf.


3) OpenAI and Cloudflare release new tools for building A-I agents

OpenAI: Updated its Agents SDK with sandboxing and harness features so developers can test long-horizon agents safely.
Cloudflare: Launched Project Think, a full-stack platform for durable agents with checkpointing, sub-agents, persistent session trees, and sandboxed code execution via Dynamic Workers.
Takeaway: Both emphasize sandboxing as essential. Cloudflare bets on a complete serverless platform; OpenAI focuses more on enterprise compatibility.


4) Gizmo raises $22 million and reaches 13 million users

What Gizmo does: Turns students’ notes into interactive, gamified study material (leaderboards, streaks, limited lives for wrong answers, friend challenges).
Numbers: $22M raised; 13M users across 120+ countries.
Why it matters: With academic performance a growing concern in the U.S., engaging learning tools have a big opportunity if they can sustain student attention.


5) Runsight launches an open-source workflow engine for agents

What it is: A self-hosted, YAML-first engine that:

  • Turns agent code into Git-native workflows.
  • Adds budget caps, pause/kill controls, and regression tests.
  • Keeps models and keys on-premise.

Audience: Free and aimed at teams that want reproducible, reviewable agent pipelines.


That’s the roundup. Stay curious, stay skeptical, and stay tuned — Isabelle and I will be back next time with more on the agent revolution and other A-I developments.

Share your love