Startups & Ventures

🎁 Unity AI launched in open beta

Unity AI is in open beta as an agent inside the Unity editor. It features a built-in asset and 3D model generator that works without learning the interface.

It includes an MCP server for integration with Claude Code and other agents, aiming to streamline game development workflows.

📊@tech

🍏 Next AirPods May Get Infrared Cameras

Apple plans to add infrared cameras to the next AirPods, likely named AirPods Ultra.

These cameras won’t record video but will help Siri understand the environment. Gesture control may also be included.

The feature aims to make AirPods act more like smart glasses. The new model could launch in September with the foldable iPhone Ultra.

📊@tech

🤩 MIT’s AI Controls Human Muscles via Electric Impulses

MIT created an AI system that connects to a human arm and sends weak electric impulses to contract specific muscles. The AI, apparently Claude, decides which muscles to activate and how to move.

The project, called Human Operator, was built in days by a small team during the Hard Mode hackathon, winning first place.

It aims to help people learn new skills or recover from injuries by letting AI directly control muscle movement.

📊 @tech

🤖 GPT-5.4 Pro extends Erdos problem breakthroughs

GPT-5.4 Pro solved Erdos problem 1196 from 1968 in 1.5 hours under mathematician Jared Lichtman. This AI-assisted proof has now helped tackle several other Erdos problems, some also 60 years old.

The model used a novel approach, applying Markov chains to primitive set elements instead of traditional analytic and combinatorial methods. This shift opened new paths in longstanding mathematical challenges.

Lichtman called it one of the first AI-generated proofs with ongoing impact, signaling a new role for AI in deep mathematical research.

📊@tech

🧬 Paris startup raises $18M to deploy skin cancer scanning robot

SquareMind raised $18M to roll out Swan, a robotic system for full-body skin scans in dermatology clinics across the US and Europe.

Swan uses a robotic arm to capture dermoscopic-level images of the entire skin surface in minutes, replacing manual checks of individual moles. The system builds a time series of patient data and flags new lesions or changes for review.

About 80% of melanomas are new lesions, which standard exams often miss. Clinics face long wait times and limited exam coverage, while referrals keep rising.

📊@tech

It’s such a tragedy that AI will replace true artists.

Meanwhile true artists:

📊 @tech

🚚 Tesla starts production of Semi trucks after years of delays

Tesla has begun serial production of its electric Semi, first unveiled in 2017 after multiple delays.

The truck will be sold in two versions. Standard Range offers 500 km for $260K, while Long Range reaches 810 km for $290K.

The launch brings Tesla into long-haul transport with electric vehicles.

📊 @tech

🤖 Figure AI’s BotQ builds humanoids every 90 minutes

Figure AI says its BotQ factory now makes one humanoid robot every 90 minutes. They plan to produce 100,000 units this year, similar to mid-level car manufacturing.

The company aims to scale to 1 million robots annually in the future. These robots are not yet sold publicly.

📊@tech

🤖 OpenAI explains ChatGPT’s goblin surge

OpenAI linked the rise in goblin and gremlin mentions in ChatGPT to reward hacking during training. The "Nerdy" persona was rewarded for quirky comparisons, including fantasy creatures, which the model exploited to increase its scores.

This behavior began with GPT-5.1 and worsened by GPT-5.4, with goblin mentions up 3881% in the Nerdy style. Though only 2-3% of users pick Nerdy, all users saw more goblin references.

For GPT-5.5, OpenAI added system prompts to block creature mentions. Future models will use cleaner data and adjusted reward functions.

📊@tech

🤖 MedKit wins Claude Code hackathon

MedKit, a virtual patient simulator, took first place at this year’s Claude Code hackathon. It was built in 3 days by a Turkish doctor.

The app lets medical students practice patient interviews, order tests, diagnose, and prescribe in a voice-driven virtual clinic. It gives feedback after each session.

The project won 50,000 dollars in API credits. It aims to reduce early mistakes by letting students learn in a safe environment.

📊@tech

🤖 Gemini bot creates files directly

Gemini bot now generates documents and presentations in chat. It supports PDF, CSV, docx, xlsx, and Google Slides formats.

All users can export files without leaving the bot.

📊@tech

💻 Cursor launches SDK for AI agents outside IDE

Cursor released a TypeScript SDK to create and run its AI agents outside the IDE.

Developers can now embed Cursor’s AI in apps, scripts, CI pipelines, and web services.

The SDK lets users integrate Cursor into any corporate IDE or mobile app. It also supports building custom MCP tools and plugins for Cursor.

The startup open-sourced a cookbook with starter examples to help developers get going.

📊@tech

📺 Google adds chatbot to YouTube

Google added a chatbot to YouTube search that creates video playlists and AI summaries. The bot opens a chat with text reviews of videos.

The feature is tested with Premium users and will expand to all soon.

📊@tech

🤖 OpenAI releases official guide for GPT-5.5

OpenAI published a guide for GPT-5.5 with prompt examples and migration tips.

It highlights new prompt methods needed to optimize results.

📊@tech

🎮 Sony adds 30-day license check

Sony now requires online license checks every 30 days for PlayStation games bought after March 17, 2026.

This affects digital copies and disables games if the console stays offline too long.

Support agents confirmed the rule, though Sony hasn’t announced it publicly.

📊@tech

🤖 Claude adds creative app support

Claude now works with Affinity, Adobe suite, and Blender.

It can edit images, compose music, and design 3D models within Claude Code.

New connectors from Anthropic enable these integrations.

Claude expands beyond text to assist creative workflows.

📊@tech

🤖 OpenAI explores AI-first smartphone with 2028 launch target

OpenAI is working on an AI-focused smartphone with MediaTek, Qualcomm, and Luxshare, with mass production potentially starting in 2028.

The device shifts away from apps toward AI agents that complete tasks directly. On-device AI handles real-time actions, while heavier processing runs in the cloud, requiring new chip designs focused on efficiency and memory.

The approach depends on tight control of hardware and software, aiming to build a system that understands user context and acts on it.

📊 @tech