Sponsored by

Welcome, AI enthusiasts

The most tedious parts of your week might soon belong to a machine that simply watched you do them. OpenAI gave Codex the ability to record a task one time and rebuild it into a skill it runs on its own. Filing expenses, requesting time off, the chores nobody fights to keep, all handed off from a single walkthrough. For now only a few people can try it, but the idea is hard to unsee. Let's dive in!

In today’s insights:

  • Codex now does the jobs you Demo

  • Claude Code Now Builds Its Own Live Web Pages

  • OpenAI Helped End a Girl's 20-Year Wait

Read time: 4 minutes

LATEST DEVELOPMENTS

Evolving AI: Run a tedious task one time and Codex can take it over from there.

Key Points:

  • The recording becomes an open skill you can read and edit before Codex runs it.

  • Recording starts and stops on your command, so nothing is captured by surprise

  • Real examples include filing expense reports and submitting time-off requests.

Details:

OpenAI's Codex tracks each step as you run a task, then saves those steps as a skill it can rerun on its own. You hit record before you start and stop when you finish. It works best on recurring tasks, the kind you run the same way each time they come up. You teach Codex by doing the work in front of it, the way you would walk a new teammate through it on their first day.

Why It Matters:

OpenAI is handing real automation to people who cannot write a macro or a script. That work normally calls for coding skills most office workers never pick up. Codex drops the bar to almost nothing. The reach is small today, yet it points to a workday where anyone can offload the parts of their job they dread.

The Ultimate Guide for Usage-Based Pricing for SaaS and AI

Implementing usage-based pricing successfully requires more than just a pricing strategy. It requires financial and operational infrastructure capable of handling dynamic pricing models, real-time usage signals, and increasingly complex monetization approaches.

In this guide, you'll learn  ⤵

  • Strategic Advantages + Implementation Guidance

  • AI Use Cases for Usage-Based Pricing

  • Insights from SaaS & AI finance leaders on overcoming challenges and maximizing UBP.

Evolving AI: Claude Code now builds its own live shareable pages straight from your session.

Key Points:

  • Claude Code turns a session into a live page, with no setup or data wiring.

  • The page updates in place at the same link every time the work changes.

  • Beta is live on Team and Enterprise, viewable only inside your org.

Details:

Artifacts draw on the full session, the codebase, the connected tools, and the reasoning Claude Code worked through. One page can hold a failing test, the error spike from a monitoring tool, and the root cause in a single view. You can ask for a dashboard you sort yourself, a walkthrough of a pull request, or an incident page that grows as you investigate. Every republish lands at the same URL with version history, so anyone can open the link and roll back to an earlier state.

Why It Matters:

Claude Code is part of a wider shift where the work an AI produces is becoming a finished deliverable on its own. For anyone who spends the day beside these tools, that quietly removes a chunk of the job, the hours spent turning what the AI did into something other people can read. What it makes now arrives ready to use.

Moda is the AI design agent with taste

Moda is an AI design product where you prompt what you need, get a complete on-brand design, and edit every element on a full canvas. 

Our viral launch hit 4.4M views in days, tens of thousands signed up, and executives at major finance and tech companies now use it.

Source: OpenAI

Evolving AI: OpenAI model helped doctors find 18 answers in cases left cold for years.

Key Points:

  • Kyra, learned the cause of her muscle disease almost 20 years after her symptoms began.

  • Researchers at Boston Children's Hospital, Harvard, and OpenAI ran 376 unsolved rare disease cases back through an o3 reasoning model.

  • After doctors reviewed and confirmed each lead in the lab, 18 patients and their families finally got a diagnosis.

Details:

Kyra's trouble started in a karate class when she was nine and could no longer hold her stances. Nearly two decades of tests gave no answer. The OpenAI model linked her case to a variant in a gene called HSPB8 and pointed to a rare muscle disease, which doctors then confirmed. By then she was 27. Seven of the 18 answers turned out to already exist in public databases, lost across records that never lined up.

Why It Matters:

Doctors at the Manton Center cannot hold 8,000 rare diseases in their heads, and the evidence behind each case keeps shifting, as genome never changes. The science around it does, so an old test can finally be read years later. A tool that helps experts revisit cold files puts answers back within reach for families who long ago stopped expecting them.

👀 Click on the image you think is real

QUICK HITS

💸 Investors Pour $1.5B Into Baseten as "AWS for Inference" Bet Pays Off.

🎮 Snap's Second Spinoff of 2026 Sends Its AI Video Team Into Gaming.

🔬 Commerce Secretary Confronts ASML Over a Possible EUV Leak to China.

🔩 Amazon Eyes Selling Trainium Outside AWS in Direct Nvidia Challenge.

📈 Trending AI Tools

  • 🤖 Lindy – The simplest way for businesses to create, manage, and share agents*.

  • ✏️ Animated Drawings - Meta's tool that turns kids' sketches into animated characters.

  • 🌐 Linnk AI - AI translator and summarizer for 50+ document formats and 150+ languages.

  • 📝 DeskribAI - AI document generator that builds professional docs from prompts.

 *partner link

Reply

Avatar

or to participate

Keep Reading