5-min read

Hello AI Enthusiasts,

AI coding agents can write more code than ever. But what happens when the tools managing that code can't keep up?

Today: OpenAI's new Git backend, Wix's AI website editor and a benchmark for everyday AI tasks.

👇 Later: Our 20-edit ChatGPT vs. Nano Banana experiment.

OpenAI built Pando, a new Git backend designed to handle the growing volume of code changes from engineers and Codex agents.

Pando reduced median push latency by roughly 80% and now serves nearly a petabyte of Git data daily.

The system was developed in just 10 weeks, with Codex agents implementing much of the work.

The bigger picture: AI isn't just helping developers write code. It's changing how the tools developers rely on are built.

👉 See how OpenAI built Pando →

🤎 Sponsored

Most GTM problems aren’t one big failure. They’re one or two weak spots quietly dragging down the rest.

The free GTM Audit scores your go-to-market across four dimensions, then sequences a 90-day roadmap for your stage.

Drop your URL. Get your score in about 10 minutes.

Get My GTM Score

Doximity launched DoseCheck to verify AI-generated medication doses. In its evaluation of nine models, at least 2.3% of dose claims were incorrect.

Our latest AIHealthTech Insider explains the findings, plus AI reading cancer slides and a pancreatic cancer vaccine result that needs context.

👉 Explore AIHealthTech Insider →

🔥 More AI news

Wix announced Studio Two, combining AI prompts, visual editing and React code.

Its Aria 2.0 agent can generate layouts and components, while creators can refine the results visually or edit the code.

Why it matters: Freelancers and agencies can build websites faster without giving up control over the design. Explore Wix Studio Two →

OpenAI introduced LifeVal, a benchmark shaped by feedback from thousands of ChatGPT users. It measures tasks such as finding childcare, accessing government services and managing personal finances.

Why it matters: Technical benchmarks don't always show how useful AI is in everyday life. Explore OpenAI's LifeVal benchmark →

SemiAnalysis released GDPx on November 2, testing AI agents on 206 difficult professional tasks. Claude Opus 5.5 led at 12%, followed by Gemini 4 Carbon at 10.9% and GPT-6 Astra at 9.6%.

Why it matters: These benchmark scores highlight the limits of agents on demanding tasks, not their overall accuracy. Compare the AI agents →

🤎 Sponsored

You don't have to scroll every AI thread, track every new tool, or watch every demo. 

The Rundown AI breaks it all down for you — the latest AI news, tools, and tutorials in one free 5-minute email every morning. 

Trusted by 2M+ professionals at Apple, Google, and NASA.

Sign up to start learning.

We tested Nano Banana 2.1 and ChatGPT Images 2.5 with identical prompts and real photographs.

ChatGPT followed several editing instructions more closely. Nano Banana often blended additions more naturally.

Both changed details they were asked to preserve.

Our Medium experiment includes all 20 edits, side-by-side images and reusable prompts.

👉 See which editor handled all 20 photo edits better →

Our latest Sunday Special explores McKinsey's take on AI value, three remote jobs and a receipt-scanning app you can build.

Also inside: November 4 webinars, cash-prize hackathons and practical AI courses.

👉 Explore the Sunday Special →

🤎 Sponsored

Top engineers at Anthropic say AI now writes 100% of their code.

Are you using AI to write yours?

These 100+ Claude Code hacks show you exactly how. Sign up for The Code and get:

Claim your free playbook

Ask an AI image editor to change a product photo's background without altering the product.

Compare it with the original. What changed unexpectedly?

Know someone building with AI? Forward this issue.

👉 Join 70,000+ readers getting practical AI news and tools →