🏆GPT‐5.2 Pro Just Broke Math

In partnership with

✨ Welcome back, Entrepreneurs

This issue AI pushes into real work. Frontier math records fall, agents get more autonomous, world models go real-time, scientific discovery speeds up, creative pipelines automate, and low-cost AI accelerates global adoption.

🔥Big AI moves

🔹GPT-5.2 Pro sets a new benchmark on FrontierMath Tier 4

OpenAI’s GPT-5.2 Pro achieved a record 31% accuracy on FrontierMath’s toughest Tier 4 problems, more than doubling the previous best score of 19%. The model solved expert-level, research-style math challenges designed by leading mathematicians, signaling rapid progress in AI’s ability to tackle frontier mathematics.

🔹Cursor ships agent upgrades with subagents, skills, and built-in image generation

Cursor’s latest update expands its agent system with parallel subagents for faster execution, Skills for task-specific workflows, and native image generation inside the editor and CLI. The release introduces AI-aware Cursor Blame, improved context handling, and interactive clarification questions, enhancing agents' power, focus, and readiness for complex codebases.

🔹Claude Code levels up with Tasks for real multi-agent projects

Anthropic replaced Todos with Tasks in Claude Code, introducing a new system for managing complex, multi-session projects. Tasks support dependencies, persist in the filesystem, sync across subagents in real time, and enable better coordination as Claude runs autonomously for longer.

🔹Anthropic brings Claude directly into Excel for smarter modeling

Anthropic introduced Claude in Excel, a new beta feature that lets users analyze, update, and debug complex spreadsheets inside Excel. Claude understands entire workbooks—including nested formulas and cross-sheet dependencies and provides clear explanations with cell-level citations.

🧠 Google DeepMind introduces D4RT for four-dimensional scene understanding

Google DeepMind launched D4RT, a unified AI model that lets machines understand the world across 3D space and time directly from video. It reconstructs dynamic scenes and tracks motion in real time using a single, efficient architecture.

Highlights

• Reconstructs moving 3D scenes from standard video

• Tracks objects across time, even through occlusion

• Combines tracking, depth, and camera pose in one model

• Runs up to 300× faster than prior methods

• Scales to real-time robotics and AR use cases

Why it matters:
D4RT pushes AI from static vision toward real-time world models, enabling safer robots, low-latency AR, and more realistic physical intelligence at scale.

The Tech newsletter for Engineers who want to stay ahead

Tech moves fast, but you're still playing catch-up?

That's exactly why 100K+ engineers working at Google, Meta, and Apple read The Code twice a week.

Here's what you get:

• Curated tech news that shapes your career - Filtered from thousands of sources so you know what's coming 6 months early.

• Practical resources you can use immediately - Real tutorials and tools that solve actual engineering problems.

• Research papers and insights decoded - We break down complex tech so you understand what matters.

All delivered twice a week in just 2 short emails.

Join 100K+ engineers

🩺 AIHealthTech Insider

🔹 Sakana AI teams up with Google to accelerate scientific discovery

Sakana AI has partnered with Google, receiving financial investment to integrate Gemini and Gemma models into its research platforms. This collaboration enhances Sakana's work with The AI Scientist and ALE-Agent, aiming to advance secure AI for scientific discovery and critical applications in Japan's finance and government sectors.

🔹What Changes When Drug Discovery Becomes AI-Native?

This issue examines how AI, robotics, and accelerated computing are being built directly into drug discovery workflows. Partnerships like NVIDIA × Lilly signal a shift from AI as an experiment to AI as core research infrastructure.

👉 Read Issue #84 to see how AI is reshaping discovery, development, and scale in biomedicine.

🎬 Create Videos, Images, and Ads Fast with Pippit by CapCut

Pippit is CapCut’s AI-powered creative agent that turns one idea into videos, images, and ads in minutes. No editing skills, no production team. Just prompt, generate, and publish -perfect for entrepreneurs and creators who need to scale content fast.

How It Works:

• Enter One Prompt → Describe your product, campaign, or idea.

• Generate Assets → Pippit creates videos, photos, designs, and ads automatically.

• Choose Formats → Marketing videos, avatars, talking photos, product shots, showcases.

• Customize Lightly → Tweak text, style, or visuals if needed.

• Publish Instantly → Export or post directly to social platforms.

Pro Tip: Use Agent Mode to spin one idea into multiple ad creatives, ideal for testing, launches, and always-on marketing.

⚙️ Sunday Special

🔹The $8 AI Era Has Begun

This issue looks at what happens when advanced AI becomes globally affordable. With ChatGPT Go expanding to 170+ countries, access, not capability is now the biggest accelerant of adoption.

👉 Read this Sunday Special to understand the implications of low-cost AI at scale.

Quick question about newsletter ads

We’re running a super short survey to see if our newsletter ads are being noticed. It takes about 20 seconds and there's just a few easy questions.

Your feedback helps us make smarter, better ads.

Take the survey now.