AI Day, Claude Impact Lab and Claude Duet

Hey 👋

Welcome back! Last week Anthropic previewed Claude Duet, an agent that keeps working while your laptop is off, and handed its skills format to the Linux Foundation.

AI is also getting scarily good at finding security holes. Last week two serious flaws in PHP, the software behind a large share of the world's websites, were made public, and researchers found them with help from AI.

Our focus in this issue is how you can tackle big company bets with AI, and Claude skills for your team. We also look back at AI Day and Claude Impact Lab at Mesh, and tomorrow we run our Claude Code bootcamp for everyone.

In this issue

• AI Day and Claude Impact Lab at Mesh
• Our bootcamp for everyone tomorrow
• A quick check of your passwords
• AI News, with Claude Duet

AI Day at Mesh

On Wednesday 21 October, Mesh Youngstorget gave the whole day to AI as part of Oslo Innovation Week, with every stage in the building in use. Aleksander hosted the day.

Every stage was full from morning to evening, and it was great to see so many people discussing both the challenges and the opportunities of AI.

Claude Impact Lab

The same day, Claude Community Oslo ran Claude Impact Lab with Mesh and us. Over 30 people in mixed teams picked a problem and built a first version with Claude Code that others could click through, without writing code. The day ended with live demos of the prototypes.

Some of the themes the teams took on

• Learning that makes kids think for themselves, instead of handing them the answer
• Making it easier for people to meet, and tackling loneliness
• Avoiding a two-tier society, where some benefit from AI and others fall behind

Each team let Claude interview them before asking for solutions. The same start works on any problem at work, and the participant guide has the prompts.

Interview me about [problem], one question at a time. Then write the problem back to me in one sentence, who it affects, what changes for them and why it matters. Wait for my go before you suggest solutions.

A good week to check your passwords

Microsoft reported last week that one group of attackers used AI agents to target hundreds of organisations. If you have not done it already, two steps make a big difference.

• Use a password manager, such as 1Password, Bitwarden or the one built into your Apple or Google account, so every account gets its own strong password
• Turn on two-factor authentication for email, work accounts and anything with payment details, preferably with an authenticator app or a passkey rather than SMS

Both take an evening to set up.

Two new guides

How to tackle big company bets with AI plans a strategy day on a big bet, with an AI agent in every group, from AI interviews before the day to clickable prototypes. It is also in Norwegian.

Claude skills for you, your team and the company shows how a skill moves from you to your team and the whole company, and why most should stay where they started. It is also in Norwegian.

Coming up

NDC Porto, 17 to 20 November. Aleksander speaks at Kopke Hotel in Porto.

We also run company bootcamps built around your team's tools, workflows and priorities. Get in touch if you want one for your team.

AI News

Here are the other updates from last week worth a look.

Claude Duet works in the background. Anthropic's new app briefs you each morning from your connected apps and finishes tasks you hand it. It asks before it sends, pays, signs or deletes anything. It is a waitlist preview for Pro and Max, for now only in the US.

Unfortunately we have not got access yet, but people we trust rate it highly. Dan Shipper at Every calls it the first personal agent he would call a pro tool.

Chat and Cowork become one Claude for Enterprise. You no longer pick a mode first. Simple questions get a chat answer, while tasks with files or code run in the cloud and keep going after you close your laptop.

Anthropic gives skills to the Linux Foundation. The skills format now sits with the Agentic AI Foundation, part of the Linux Foundation, so no single company owns it. Anthropic says more than 70 tools support it, including ChatGPT, GitHub Copilot and Cursor. A skill your team writes today is not locked to one tool.

ChatGPT can lock chats. You can hide chosen conversations from your history and search, and open them with a PIN or your face or fingerprint. It is rolling out gradually, on every plan from Free to Enterprise.

Interactive slides in PowerPoint. Describe what you want, and Copilot builds a slide your audience can click through, such as a timeline or a dashboard. Microsoft says it reaches PowerPoint on Windows, Mac and the web by the end of October, for people with a Copilot licence.

AI helps find critical security flaws in PHP. Security researchers used AI models, including OpenAI's, to find the two critical flaws in PHP. The updates are out, and Norway's security authority NSM advises installing them as soon as possible.

The tools available today can do far more than most people know is possible. For builders and problem solvers, the real limit now is imagination.

If you know someone who should join us tomorrow, feel free to forward this email.

Best, Aleksander and Simon