Welcome to Jumble, your go-to source for AI news updates. A breakthrough retinal implant lets people who lost central vision read letters and words again, while Anthropic’s Claude Code lands on the web with agent style workflows you can run from a tab. Let’s dive in ⬇️

In today’s newsletter:
📖 New implant restores reading for blind eyes
🧑‍💻 Claude Code arrives in a browser
🌏 China’s AI users climb to 515 million
📺 Channel 4 tries an AI presenter
🗓️ Weekly Challenge: Connect your calendar to ChatGPT

🤓 AI-Powered Implant Helps the Blind Read Again

In a European trial led by Moorfields and University College London, surgeons implanted a two-millimeter microchip called PRIMA beneath the retina of people who had completely lost central vision from dry age-related macular degeneration (AMD). 

🧠 How the System Works

The device works with camera glasses and a pocket-sized processor to rebuild a digital bridge between light and sight. The glasses capture the world in front of the wearer and send it to the belt-worn computer. There, AI-powered vision software analyses what it sees — detecting letters, contrast, and edges — and translates that data into infrared light patterns. 

Those patterns are projected back into the eye, where the PRIMA chip converts them into electrical signals that stimulate surviving retinal cells. The brain interprets those pulses as visual shapes and text, producing what patients describe as “a new way of seeing.”

👁️ Results From the Trial

The study involved 38 participants across six European centers, including five treated at Moorfields. Before surgery, none could read at all. After the implant was activated, 84% regained the ability to recognize letters or short words, with some reading up to five lines on a standard eye chart. 

One participant said the moment she spotted her first letter felt “dead exciting” — like learning an entirely new language for the eyes.

⚙️ Why It Matters

Dry AMD, especially its advanced form known as geographic atrophy, currently has no approved treatment. The PRIMA system offers the first realistic path toward restoring functional central vision rather than just slowing decline. Here’s what Neuroscientist had to say about it earlier this year:

The surgery itself takes under two hours and doesn’t require any internal wiring or batteries; the implant is powered optically through the glasses and remains dormant when not in use.

🔮 What’s Next

The PRIMA system was originally developed by Pixium Vision and is now being advanced by Science Corporation. Researchers plan smaller processors, faster AI rendering, and sharper resolution for future versions. 

Credit: Moorfields Eye Hospital NHS Foundation Trust

Regulators will be watching long-term safety and real-world performance — but for patients once locked out of books, menus, and crosswords, this tiny chip marks a monumental step toward independence.

🧑‍💻 Anthropic Unveils Claude Code on the Web

Anthropic just brought Claude Code to the browser, turning what used to be a developer-only tool into something anyone can try. Think of it as a studio for automated coding tasks that runs safely in the cloud. 

You open a Code tab, describe what you want—“clean this script,” “add a chart,” “test this workflow”—and Claude does the heavy lifting on Anthropic’s servers. You can queue several jobs, track their progress, and review the results from one simple dashboard.

🧠 Why Non-Coders Should Care

Instead of local installs or terminal commands, everything happens in a secure browser space. That means no setup headaches, no tangled dependencies, and no risk to your own files. It’s built for people who use tech every day but don’t live inside GitHub.

Claude Code is part automation assistant, part coding coach. You can feed it a CSV and ask for a quick data summary, turn an idea into a working prototype, or generate snippets that automate chores you’d normally outsource to IT. The tool’s AI-driven reasoning helps translate plain English instructions into clean, executable logic, then shows you what it changed line by line.

🛑 Limits and Risks

Claude Code is still in research preview, so expect guardrails. It can’t yet handle full-scale deployments, and anything that touches live systems should be reviewed by a human. Like all coding AIs, it sometimes invents missing functions or leaves gaps that need cleanup. Some users also note potential safety risks.

But the early signs are strong: structured task management, clear explanations, and a calm web interface that makes complex automation feel as easy as describing what you want done. For tech-savvy users, that’s a major step toward hands-on AI productivity.

This Week’s Scoop 🍦

🧩 Weekly Challenge: Connect ChatGPT to Your Calendar

Challenge: Link your calendar so ChatGPT can find openings, summarize your week, and draft holds for you.

Here’s what to do:

  1. In ChatGPT, open Settings then Connectors. Choose Google Calendar or Outlook Calendar and select Connect. Approve the read level permissions. You can toggle automatic use in chat anytime.

  2. Ask simple checks: “What are my free windows on Friday afternoon” or “List my next three meetings with locations.” ChatGPT reads events and replies inline.

  3. Try a workflow: “Find a one hour slot next week with no conflicts, draft an agenda called Q4 plan, and create a prep checklist.” If you connected email too, ask it to draft a short invite you can paste into your calendar.

⚠️ Safety tip: Only grant the minimum access you need and disable automatic use if you prefer explicit prompts. You can disconnect connectors in Settings at any time.

Want to sponsor Jumble?

Click below ⬇️

If the AI revolution is giving sight to the blind, what else will it do next? And, do you think Claude Code on the Web is a game-changer, or just another AI tool we’ll forget about? We’d love to hear your thoughts. See you next time! 🚀

Stay informed, stay curious, and stay ahead with Jumble!

Zoe from Jumble

Keep Reading

No posts found