This week in tech, Google claims the top spot for AI models, Meta's AI agent muscles into small business software, and a university research team tackles one of AI 3D printing's most annoying problems, meshes that look great on screen and fall apart on the plate. Here's what actually matters if you make things for a living.
Google's Gemini 4 Argon Claims the Top Spot
Google released Gemini 4 Argon, a new model it says outperforms rivals from OpenAI and Anthropic on coding, research, and cybersecurity benchmarks. For now it's only rolling out to select cybersecurity partners, with a wider release expected later. For a small creative business, this matters less for the cybersecurity angle and more for what trickles down: every time a flagship model gets sharper at coding and writing, the consumer tools built on top of it (product descriptions, email replies, quick mockup copy) tend to get noticeably better within a few weeks.
Meta's AI Agent Muse Comes for Small Business
Meta expanded its AI agent Muse to small businesses, hooking it into tools like Shopify, Slack, and Dropbox so it can look across sales, inventory, and campaigns instead of answering one question at a time like a standard chatbot. It's free to start, with paid tiers for more. If you're running a one-person studio, the appeal is obvious: less tab-switching, one agent that can actually see your store and your calendar at once. Whatever AI layer you bolt on top, it still needs a clean, professional home base underneath it. That's usually the deciding factor between a shop that looks like a hobby and one that looks like a business.
A New Tool Fixes Broken AI 3D Models Before You Print Them
Researchers at MIT, Google, and Northeastern built InstructMesh, a tool that pairs image generation with language models so someone with zero CAD experience can spot and fix structural flaws in an AI-generated 3D model before sending it to a printer. In testing, novices caught and corrected the problems roughly 90 percent of the time, turning shaky AI meshes into mugs, glasses, and even small robots that actually held together. Anyone who has typed a prompt into an AI 3D generator and gotten back a mesh that looks perfect on screen and prints like a collapsed sandcastle knows exactly why this matters. If you want to try that text-to-3D pipeline today instead of waiting for the research to ship, Tripo 3D is the AI 3D generator worth a look for turning a prompt or reference photo into a usable mesh.
OpenAI's Dots Is an Always-On Agent for Busywork
OpenAI launched Dots, an assistant built on its GPT-6 Astra model that runs in the background to chase down a goal without needing constant prompting, available to ChatGPT Pro and Business Premium subscribers through places like Slack and Teams. For a small creative business, the busywork that eats a day, following up on a lead, nudging a customer for a reference photo, drafting a quick reply, is exactly the kind of task these background agents are built for. It's early and currently locked behind the higher-priced plans, but worth watching as access and pricing widen.
That's Tech Thursday
That's the tech and AI news that actually matters for creators this week. Want it turned into something physical? That's what the studio does. See you next Thursday.
Some links in this post are affiliate links. Prices stay the same for you, I earn a small commission if you buy through them.