As a reminder, each week I cover:

🏕️ Summer camp kicks off. The INSEAD AI Venture Lab is underway, with free building tools, a $50k grant pot, and a "record label for founders" business model

🤖 What actually is an AI skill? I learned more this week

📚 Warts and all founder book series. No ‘playbook’, just journals from failure to $1m in profit

🏃 Parkrun, MyWhoosh, and why fitness is on the schedule this week

This week my AI Venture Lab programme with INSEAD got going properly. I attended a technical onboarding to the Audos platform with co-founder Nicholas Thorne.

As a participant, I can build my project entirely for free — the technology is powered by Anthropic, with Cursor handling the vibe-coding side, which saves me real money. There's also a $50k grant pot I can compete for, plus a solid programme of workshops and group mentoring over the coming months.

The interesting part comes at the end. If I build something with revenue potential and want to stay on Audos, they take a 10% cut of sales. They pitch themselves as a "record label for founders", which is a genuinely distinctive model. If I get there, it'll be a significant decision, likely depending on how much value Audos has added and what it continues to offer. I'm free to take what I've built elsewhere if I choose not to stay.

Honestly, if I'm facing that dilemma, it'll be a great problem to have!

If you’re interested to find out more about Audos, this short video gives an idea,

Reminder: I'm looking for people who love discovering new films, art, music and literature — sign up for the free Polytoto waitlist at Polytoto.com

This week's rabbit hole started with a viral post about something called a "grill-me" skill — a tool one developer built that interrogates him relentlessly about a plan until they've reached a shared understanding, then produces a summary.

My first reaction: this just sounds like a good prompt. Why does it need to be a special file?

So I looked into it.

It turns out the answer isn't about the words themselves; it's about what happens before you type anything. A skill is a folder with an instructions file that sits dormant in the background of every conversation. When a session starts, the AI quietly reads a short description of each skill it has access to, just enough to know when it's relevant, without loading the full thing. The moment you say something like "grill me on this," it recognises the match and pulls in the complete instructions automatically. No copy-pasting from a notes doc. No remembering it exists. It just fires when it's needed.

That's the key benefit: a skill is self-triggering. A prompt you paste in is something you have to carry around and remember to use. A skill is closer to installing a small piece of software: reusable, shareable, always on.

For anyone building a repeatable workflow, it's worth considering. This newsletter is a good example: tasks like proofing and bullet-point suggestions recur every week and could be automated. It also reminded me of a chapter in The Complete Founder on process management.

Documenting repetitive tasks so that someone else (or an AI) can handle them is entrepreneurial efficiency.

One of the most honest founder stories I've come across recently comes from Alex West, a software entrepreneur who has published a series of free books documenting his journey from a string of unsuccessful side projects to eventually generating over $1m in profit from his own products.

What makes the series unusual is that it isn't written as a playbook or a collection of productivity hacks. It's a chronological record of experiments, setbacks, doubts, and occasional breakthroughs, drawn from daily journals kept over several years.

For founders, the value lies less in the financial milestones than in the mindset. West shows repeatedly that entrepreneurial capability isn't built by consuming more advice — it's built by building, testing, reflecting, and trying again.

His early books are full of failed launches, abandoned ideas, and wrong assumptions, yet each attempt leaves him slightly more capable than before. It's a useful reminder that upskilling as a founder is not about finding the right framework; it's about accumulating experience, developing judgement, and getting a little better with every project.

Until now, my sabbatical has been exploratory and self-directed — I've followed my own curiosity. That changes a little, at least until the end of the INSEAD summer camp. Building the Polytoto prototype means committing to a lane and doing what the project demands. That includes continued AI development, and this week I'll be deep in app-building.

On Founderval, I've drafted the structure of a one-hour interactive workshop on entrepreneurial skills development, and this week I'm completing the slides. I like using the visual design process as a way to evolve content — thinking about what I'm communicating and how it benefits the learner as I work through each slide.

I'm also stepping up my fitness this week. The Complete Founder covers Bandura's four sources of self-efficacy, one of which is emotional state and physical wellbeing. I completed my 26th Parkrun of the year on Saturday, but holiday indulgence has hit my times — so a few sessions on MyWhoosh are going in the diary, riding the 3D landscapes of San Francisco and Mount Fuji.

Have a good week!

Luke

🎩Hat-Tip This ad for a luxury goods resale platform just won a creativity prize at Cannes. It's entirely AI-generated. Meanwhile, Purga Studios has introduced itself as an AI-only film production company, with the tagline: "Lorem Ipsum: We have the craft. We need the scripts."

LinkedIn | YouTube | Instagram | TikTok