This Week in Tech 125

What we explore this week:

Google's I/O conference this week looked like a coordinated assault on OpenAI's AI leadership. Google DeepMind launched Gemini Omni, their answer to OpenAI's Sora that can generate and edit video from any combination of text, images, audio, or sketches. Meanwhile, Gemini 3.5 Flash targets AI agents and coding with frontier-level performance at 4x the speed and half the cost of competitors.

The real ambition shows in Google's platform plays. Search will soon let you deploy multiple specialized AI agents simultaneously, turning Google's search engine into a command center for managing autonomous assistants that crawl the entire web in real-time. For researchers, Gemini for Science promises to accelerate discovery by helping explore hypotheses, validate experiments, and digest literature at machine speed.

Even development itself is going mobile: Google AI Studio is coming to phones, potentially democratizing AI development beyond desktop-bound developers. This feels like Google leveraging its massive compute advantage to make sophisticated AI accessible at scale while OpenAI focuses on flagship models.

When one of AI's most respected voices leaves OpenAI and joins Anthropic, it's not just a career move—it's a statement about where the field is heading. Karpathy's announcement that he's joining the Claude maker suggests the race for responsible AI leadership is becoming as important as the race for raw capability.

For context: Karpathy basically created the modern playbook for training neural networks. He built Tesla's self-driving AI system, then helped lead OpenAI through its breakthrough GPT era. When someone with his track record calls the next few years "especially formative" and chooses the company known for AI safety research over the one racing toward artificial general intelligence, that's a signal about which approach might win long-term.

This move positions Anthropic as the destination for top AI talent who want to build powerful systems responsibly, while OpenAI increasingly looks like the move-fast-and-break-things option in the space.

When you type a prompt, you summarize. When you speak one, you explain. Wispr Flow captures your full reasoning — constraints, edge cases, examples, tone — and turns it into clean, structured text you paste into ChatGPT, Claude, or any AI tool. The difference shows up immediately. More context in, fewer follow-ups out.

89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Try Wispr Flow free — works on Mac, Windows, and iPhone.

Start flowing free

Most AI assistants are glorified search engines that talk back to you. Higgsfield is launching what they claim is the first cloud-native AI agent that actually completes end-to-end tasks with 40+ built-in tools and persistent memory, accessible through browsers or Telegram.

Meanwhile, Tempo claims to be the first AI that can run growth marketing end-to-end without human oversight, handling strategy, execution, and optimization autonomously. If these agents work as advertised, we're looking at the shift from AI chat toys to AI that actually gets stuff done.

The most dramatic example: Google's Antigravity 2.0 built a working operating system in 12 hours using 93 AI agents for under $1,000 in compute costs. The team even booted Doom on the AI-built OS during their demo. When AI can architect entire systems instead of just writing code snippets, software creation becomes as fast as having an idea.

OpenAI launched personal finance features in ChatGPT Pro, letting users connect bank accounts via Plaid to get spending insights and ask AI questions about their transactions. This isn't just another AI feature—it's OpenAI moving directly into territory that hundreds of fintech startups have been trying to own, instantly making their AI-powered budgeting and financial analysis tools look obsolete.

The move shows how frontier AI companies can instantly enter new markets just by adding features, making specialized startups vulnerable to platform players who already have the user base and AI infrastructure.

The spatial computing wars are heating up as Google and Samsung prepare to launch their smart glasses this fall, directly targeting Meta's Ray-Ban partnerships and Quest ecosystem. This marks Google's return to the AR glasses market after Google Glass flopped a decade ago, but now they're betting Samsung's hardware expertise can make the difference.

Google's approach focuses on audio-first functionality powered by Gemini AI instead of trying to cram screens into your face. These new specs prioritize practical AI assistance through sound: photo capture, real-time translation, and navigation without looking at your phone. It's a much smarter approach that could actually stick this time.

The timing coincides with Android XR finally getting real hardware in 2026, Google's most serious attempt yet to challenge Apple's spatial computing ambitions. With Apple Vision Pro struggling and Meta dominating, this could reshape who controls our face computers.

After years of clunky AR headsets, ASUS ROG and XREAL just dropped the R1 with 240Hz refresh rates and a virtual 171-inch display that actually sounds built for serious gaming. This isn't another gimmicky AR gadget—it's positioning itself as the first real alternative to traditional gaming monitors for competitive players who want massive screens without the desk space.

Speaking of practical AR applications, Ray-Ban's smart glasses ecosystem continues expanding beyond basic photo capture. Developers have created everything from playable Doom and brick breaker games to smart home controls, all running directly on the glasses. These practical use cases suggest AR's breakthrough moment might come through simple, useful applications rather than flashy metaverse promises.

Matrix Robotics just dropped their MATRIX-3 humanoid robot at $99K, finally breaking into territory where businesses might actually buy these things instead of just demo them. At 170cm tall with 27-degree-of-freedom hands and human-like walking speed, this isn't just another research project but a genuine attempt at commercial humanoid robotics.

The real question isn't the tech specs—it's whether we're ready for human-sized robots walking around workplaces. The sub-$100K threshold feels like the moment humanoid robots transition from science fiction to business equipment, similar to how industrial robot arms became standard factory tools once the economics made sense.

Google's I/O conference this week felt like watching a tech giant finally coordinate all its superpowers into a single offensive. Instead of scattered AI experiments, we saw a cohesive vision: Gemini models that work across every surface from search to smart glasses, agents that actually execute tasks instead of just chatting about them, and development tools that put AI creation in everyone's pocket. It's the kind of platform play that could reshape entire industries overnight.

But the most telling moment might have been Andrej Karpathy's quiet announcement that he's joining Anthropic. When one of AI's most respected architects chooses the safety-focused lab over the AGI racing team, it signals something important about where the smart money thinks this field is heading. The companies that figure out how to build powerful AI responsibly might have more sustainable advantages than those just chasing the capability frontier.

Meanwhile, we're finally seeing the pieces come together for ambient computing that actually works. Sub-$100K humanoid robots make commercial deployment realistic. Audio-first smart glasses solve the social awkwardness problem that killed Google Glass. AI agents that can spin up entire operating systems in hours suggest we're entering an era where software creation becomes as fast as imagination. The question isn't whether these technologies will work—it's whether we're prepared for how quickly they'll change everything.

The next few years really will be especially formative, as Karpathy put it. The winners won't just be the companies with the best models, but the ones that can make powerful AI feel natural, useful, and safe in our daily lives.

Kingscrowd Capital's senior analyst reviewed RISE Robotics and projected potential growth to a $500 million valuation. The community round is open now on Wefunder. You don't have to be an institutional investor to get in at today's price.

See the Opportunity

Your designer is busy. Agencies are expensive. And the deck needs to go out today.

Moda is your AI design partner for the moments when you need great work fast.

Drop in your brand, tell Moda what you need, and get polished, on-brand slides, docs, or ad creative in minutes.

Try Moda Free Today