Google’s out-of-this-world AI

Plus: Amazon battles Perplexity


Welcome back! OpenAI just released Sora on Android, expanding its AI video creation app to millions more users. Now anyone can generate videos, remix content, and create digital cameos directly from their phones.

What happens when everyone can make a movie scene in under a minute? I’m eager to find out. 🍿 

Google’s Plan for ‘Data Centers in Space’

Google will tap satellite company Planet Labs to help build the Suncatcher hardware. Photo via Planet Labs.

Google thinks the future of AI might not be on planet Earth. The company announced Project Suncatcher, a research effort exploring whether its AI chips could operate on solar-powered satellites, effectively creating data centers in orbit.

How it would work: Google imagines constellations of satellites equipped with its Tensor Processing Units, powered by uninterrupted sunlight, communicating with each other in orbit. The company says solar panels in space could be up to eight times more productive than on Earth. Plus, access to near-continuous solar energy would bypass the growing strain AI is placing on terrestrial power grids.

There are a handful of major hurdles: 

  • Satellites would need ultra-high bandwidth to replace ground data centers.

  • Chips must survive radiation long term.

  • Launch and deployment costs remain extremely high.

So? Even Google admits this is a mid-2030s possibility at best.

Why it matters: AI growth is now limited by energy, not just hardware or algorithms. Google exploring space-based compute signals how serious the resource question has become, and how far (literally) companies are willing to go to solve it.

Agentic Browsing: Amazon vs. Perplexity

Amazon issued a legal notice to Perplexity regarding the use of its AI browsing and shopping agent, Comet, on Amazon’s retail site. 

Breaking down the drama ➡️ 

  • Amazon states that Comet functions as software-based automated retrieval, regardless of whether a user directs the requests. So, it must disclose itself as an agent and comply with commercial access rules.

  • Perplexity argues that Comet performs user-initiated actions, similar to how a browser extension or voice assistant interacts with websites, and should be treated as a user interface, not a bot.

The bigger picture: Major online platforms have not yet standardized how AI agents acting on behalf of users should be classified (whether as users, apps, or automated scrapers). The outcome of this dispute could set the standard for how AI assistants are allowed to navigate digital platforms.

Former Meta Researchers Launch Smart Voice-Input Ring

A team of former Meta Reality Labs and CTRL-Labs researchers has released Stream, a smart ring designed for voice input and gesture-based media control. Think: capturing short thoughts or adjusting audio, all from your finger.

How the device works: 

  • Stream includes a capacitive touch sensor used to start and stop recording. When activated, its directional microphones pick up close-range, low-volume speech, even whisper input. 

  • Recorded audio is transcribed in the companion app, where users can search or query their notes using integrated AI models. 

  • The ring also supports gesture controls for play/pause, track changes, and volume adjustments.

Why the design matters: Stream reflects an approach focused on capturing transient thoughts and supporting quick, situational interactions. It’s less about replacing smartphones or other devices and more about augmenting them.

One Chat. All Your Tools. All the Context.

Message threads and group chats have been just chatter—until now. Glue, the first multiplayer MCP client built for developers and AI‑first teams, transforms it into a multiplayer control panel.

Glue gives you:

  • Explicit and transparent tool calls—so no more guessing who triggered what.

  • Seamless workflows—run Linear, Notion, Vercel, Asana, and more all from inside your conversation.

  • Log tickets, ship updates, or manage docs without ever leaving the thread—close those tabs.

Ready to stop settling for channel overload or lost context?

AI-powered note-taking workspace

Via Chopdi

Chopdi auto-tags notes, generates summaries, and lets you chat with your entire knowledge base.

How you can use it:

  • Turn messy meeting notes into polished outlines

  • Search ideas instead of exact keywords

  • Auto-generate summaries and tasks

  • Export clean PDFs instantly

Pricing: Free and paid plans available

Create videos from text

Via Keevx

Keevx is an AI avatar video generator that turns a written script into a lifelike, lip-synced video in seconds.

How you can use it: 

  • Make training videos without filming

  • Clone your own avatar for branded content

  • Produce multilingual marketing campaigns

Pricing: Free and paid plans available

Jobs, announcements, and big ideas

  • OpenAI faces seven lawsuits over ChatGPT and mental health issues.

  • Amazon introduces Kindle Translate, an AI-powered translation service for eBooks.

  • Google Finance adds AI-driven deep search, prediction markets data, and live earnings features.

  • AI is blamed in part for the worst October job cuts in years.

  • The EU considers delaying parts of the Artificial Intelligence Act.

  • Booz Allen Hamilton is hiring an AI engineer to design and implement AI solution architecture.

  • Xometry is looking for a senior manager to build and lead next-generation organic search and discoverability strategy.

TechCrunch Disrupt just hit San Francisco for its 20th anniversary and this year, the AI startups absolutely stole the show. From cooking robots and AI weed killers to shrimp nanotech and drone defense systems, I explored the wildest innovations shaping the future of technology.

That’s a wrap! Have a great weekend and I’ll see you back here Wednesday.

—Matt (FutureTools.io)