Damus

Recent Notes

Andy David profile picture
This week on @Good Stuff Podcast, @Pete Winn πŸ”† and I chatted about why companies shouldn't be quick to outsource AI to consultants or software vendors.

We recorded this right after a 3.5 hour workshop where about 30 non-technical people built and deployed their own web and mobile apps, so it seemed like the right time to explore this.

We’re seeing a lot of organisations trying to figure out how to create an AI strategy, but for many of them, I think this is the wrong place to start.

There's a big gap in first-hand experience with AI and this is where strategy tends to break down.

It's just really hard to create a strategy for something you don't deeply understand.

It's a bit like being asked to create a strategy for magic.

Brandon Sanderson wrote his Laws of Magic as a guide for fantasy authors. If magic has no rules and can solve any problem arbitrarily, readers feel cheated. But if magic operates within clear constraints that readers understand, then solutions feel satisfying.

AI probably looks a bit like magic to many businesses right now.

When people inside a business start experimenting with the tools themselves, rather than outsourcing AI to consultants or software vendors, the organisation makes much better decisions about where and how AI fits inside the business.

This is why developing internal capability should come before strategy.

Capability means three things.

➑️ It’s shared mental models about what AI can and can't do led by internal AI champions - people who understand the technology, adapt as the technology evolves, can answer questions based on experience, and can evangelise organically.

➑️ A safe environment where this capability can be nurtured and supported over time, alongside others facing similar challenges, where gains are methodical and compound over time, and

➑️ A way to deploy what is built directly into the organisation so that capability translates into productivity across the organisation.

That’s a flywheel for capability development. It means you own your strategy and allow it to organically form and evolve over time.

https://youtu.be/b9I28wnFiQ8
2
Pete Winn πŸ”† · 1d
nostr:nprofile1qqsg2a2r7htuwutam3t2ftcy92dypxllv65t6wj3qrctfz9lygm9wnq38prnm this is the sort of post you could rebroadcast, might be worth having a read through our transcripts
TheBitSmith · 1d
Living the β€œyou can just do things” attitude, love it! Sign me up for any virtual versions of the workshop.
JeffG · 3w
Tell me more! What are you using it for? What tool stack are you using?
Andy David profile picture
Mostly Claude Code, some Codex and Wingman - built by @Pete Winn πŸ”† which we use at @Other Stuff. I started out with Cursor and I'll still review code there, but it's predominantly in the terminal

Mostly building business tools, some work on graph databases, websites, games, and even a personal finance/tax prep tool. A lot of audience of one stuff.

Posted this one earlier, it's a Miro style infinite canvas synced with Claude Code/Wingman so we can collab in a visual space.

https://primal.net/e/nevent1qqspqtxr7z58r3mevl8e83tnhzkreeaqds3q7yjttcr76s46s9wj7ks96umgr

2❀️1
JeffG · 3w
Nice!!
JeffG · 3w
Is optikon an app you guys built? Looks really cool.
Andy David profile picture
It looks like Miro, acts like Miro, but its not Miro. It's Optikon, an infinite canvas that syncs with AI agents like Wingman / Claude Code, so we can collaborate visually in real time.

Uses Nostr for auth and identity, authorship, and collaboration scoped to pubkeys.

Here's a really rough early attempt from Claude Code mapping out how it might implement a JSON pipeline to collaborate with me visually on the canvas

22πŸ‘€1
Pete Winn πŸ”† · 3w
Big plans 🦫
JeffG · 3w
Tell me more! What are you using it for? What tool stack are you using?