Welcome, humans.

France just dropped a robot that looks like it escaped from a Kojima game. Meet Aru, a shape-shifting inspection bot from Nio Robotics that climbs stairs, opens cabinets, and contorts into tight spaces.

The design is genuinely clever: instead of trying to make a humanoid robot navigate stairs (which is still really hard), Aru uses a centaur-style body that can adjust its wheel height independently. Think Mars rover meets industrial inspection drone, but it can chase you up stairs.

One Reddit commenter: “I'm not entirely sure the advantages over a wheeled doggie, but cool nonetheless.“ The answer? Stability and payload capacity. Centaur bots can carry more weight and won't tip over when reaching into overhead compartments.

The dramatic pizzazz in the demo video is a bit much for a robot that's essentially checking valve gauges, but hey, this is France. They do cinema.

Don’t forget: Check out our podcast, The Neuron: AI Explained on Spotify, Apple Podcasts, and YouTube — new episodes air every week on Tuesdays after 2pm PST!

So ChatGPT Just Became an App Store… Here's What Developers Need to Know

So, ICMYI, developers can now submit apps to be hosted inside ChatGPT that extend conversations with real functionality. These aren’t your ma or pa’s GPTs… these are real applications for ordering groceries, building slide decks, booking apartments, all without leaving the chat.

Here's the setup: OpenAI launched the Apps SDK earlier this year at DevDay. Now they've opened the floodgates for app submissions and introduced an app directory where users can browse and search for published apps right inside ChatGPT.

• Apps work by bringing new context into conversations and letting users take action.

• Think of them as mini-programs that ChatGPT can trigger when you mention them by name (with an @ symbol) or when the AI thinks they'd be helpful based on what you're talking about.

• Developers build apps using the Apps SDK, which runs on something called the Model Context Protocol (MCP), which is a framework that lets your app talk to ChatGPT and display custom interfaces directly in the chat window.

What you can build:

• Shopping tools that pull in product catalogs and handle checkouts.

• Productivity apps that connect to your calendar, tasks, or project management.

• Data visualization widgets that turn spreadsheets into interactive charts.

• Booking systems for restaurants, travel, or appointments.

OpenAI's provided serious resources to help developers get started. There's a quickstart guide, open-source example apps, and even an open-sourced UI library for building chat-native interfaces.

The catch: Not everything flies. Apps have to follow strict submission guidelines covering safety, privacy, and quality. You also can't sell digital products yet (physical goods only), gambling and adult content are banned, and your app needs clear value beyond what ChatGPT already does natively (but be warned, if it’s popular enough, ChatGPT might ~expand~ what it does natively).

The rollout plan: Approved apps will start appearing gradually in the new year. OpenAI's vetting each submission for quality and safety. Apps that users love will get featured placement in the directory or recommended by ChatGPT automatically.

Why it matters: This transforms ChatGPT from a conversational AI into a platform where intent meets action. Instead of ChatGPT just telling you about options, apps let you actually do the thing you're talking about.

What's changing: Expect model developers and third-party builders to rush into this space over the next 6-12 months. If you're building AI tools, this is your distribution channel—600,000+ engaged users already reading about AI daily through newsletters like ours.

What you should do: If you're a developer, start researching use cases and building your MCP server now. The apps that ship early will have first-mover advantage as ChatGPT starts recommending them in conversations. You can deploy your app and submit it through the OpenAI Platform once it's ready.

The app directory is already live at chatgpt.com/apps, and the race to build the most useful ChatGPT apps just started. And if you aren’t gonna make your own? You can still go to the apps page and check out all the cool tools now available to you inside GPT.

FROM OUR PARTNERS

Agents that don’t suck

Are your agents working? Most agents never reach production. 

Agent Bricks helps you build high-quality agents grounded in your data. We mean “high-quality” in the practical sense: accurate, reliable and built for your workflows. 

Generic benchmarks don’t cut it. Agent Bricks measures performance on the tasks that matter to your business. 

Evaluate agents automatically, and keep improving accuracy with human feedback. With research-backed techniques for building, evaluating and optimizing, you can turn your business data into production agents faster — with governance built in from day one.

See how Agent Bricks works

Prompt Tip of the Day

Don’t ask “which is better?” Ask “which is better for me?”

Most comparisons turn into a spec sheet dump. This prompt forces the model to do the thing you actually need: map the options to your situation, surface risks + tradeoffs, and end with a decision rule you can reuse later.

“Compare and decide” (buy/build/choose) Prompt 

Compare [Option A] vs [Option B] for [my situation]. 
Output a table with: features, tradeoffs, risks, cost/time, best-fit. 
Then give a recommendation with clear reasoning and “if X, choose A; if Y, choose B.”
Research and show any hidden costs.
Include a short ‘What would make me regret this?’ section for each option, and end with a 3-step next action plan (trial, pilot, or due diligence).
Give a confidence level regarding how likely it is to be the right choice given what you know about my needs today and me as a person.

Treats to Try

*Asterisk = from our partners (only the first one!). Advertise to 600K readers here!

1. Caffeine.AI lets you describe an app in chat, then it generates and updates the full website/app for you so you can ship without writing code.

2. Grok Voice Agent API (via LiveKit) lets you build a low-latency, two-way voice agent so people can talk to your app naturally and get spoken responses back instantly.

3. Magistral-Small-2509 gives you a capable “reasoning brain” you can run on your own machine.

4. Moondream 3 preview lets you point it at an image and ask practical questions (count objects, identify items, locate things) so you can build visual features.

5. NVIDIA Nemotron 3 gives you open models (Nano/Super/Ultra) plus datasets and libraries so you can build multi-step “agent” workflows you can actually inspect, customize, and run yourself.

6. Sprinto keeps your audit evidence and compliance checks running without spreadsheets.

Around the Horn

1. Small language models argued the next wave isn’t just bigger LLMs—it’s efficient SLMs that run anywhere, plus SLMs that act as LLM copilots. fetch-blocked

2. Agent framework study mined 11,910 discussions across 10 frameworks to map where devs struggle most (selection, learning cost, perf tuning, maintainability).

3. MoE RL routers proposed “Rollout Routing Replay” to stop MoE RL collapse by replaying inference-time routing during training.

4. INTELLECT-3 claimed frontier-style RL at scale is becoming reproducible: a 106B MoE (12B active) trained with open RL infrastructure up to 512 H200s.

5. Pass@k Training proposed training with Pass@k rewards (plus an analytic advantage) to push RLVR models toward better exploration.

FROM OUR PARTNERS

Ideas move fast; typing slows them down.

Wispr Flow flips the script by turning your speech into clean, final-draft writing across email, Slack, and docs.

It matches your tone, handles punctuation and lists, and adapts to how you work on Mac, Windows, and iPhone.

No start-stop fixing, no reformatting, just thought-to-text that keeps pace with you. When writing stops being a bottleneck, work flows.

Give your hands a break ➜ start flowing for free today.

Tuesday Tool Tip

Google's Opal is the probably the simplest AI no-code creation tool we've tried… and it just got way easier to access.

Google just integrated Opal directly into Gemini's web app. Here’s how it works: You describe what you want to build in plain English, and Opal creates a functional mini-app in minutes. Budget tracker? Meeting scheduler? Content generator? Just type it out.

Here's what makes it dead simple: Opal shows you exactly how it interpreted your request with a visual step-by-step breakdown. Each step is editable; if one piece doesn't work right, tweak that step instead of starting over. No programming knowledge required.

Here’s how you access it: Find it in your Gems manager inside Gemini. Once you build a mini app, it's saved as a reusable Gem you can run anytime. Think of Gems as customized chatbots, but Opal-powered Gems are actual functional tools with their own interfaces.

Need more control? Switch to the Advanced Editor for granular customization. But most workflows won't need it; the in-Gemini version handles the basics beautifully.

If you already use Gemini, you already have access. Try it in Gemini now.

A Cat’s Commentary

P.S: Before you go… have you subscribed to our YouTube Channel? If not, could you?

P.P.S: Love the newsletter, but only want to get it once per week? Don’t unsubscribe—update your preferences here.