Six months ago, once I started using Open Claw, I got seriously into automation. While building various things, I remembered something I’d found lacking back when I was working, and thought it’d be convenient to have this kind of thing organized and delivered every morning, so I built one.
At first I just handed the whole thing over, but the results weren’t satisfying, so I switched models a few times after that. Using only cloud models made the token cost a burden; using only local models (Qwen, Gemma) left the quality lacking. Now I use a mix of both. The local models handle the volume, and the cloud model judges what’s actually important.
Honestly, the briefing itself might not be a big deal. Give a model one sample and ask it to make something similar, and today’s models will churn it out quickly. What was actually interesting came next: making it run every day, having it notice when it’s wrong, and fixing it based on feedback. This part was far harder than building it, and I learned a lot from it. These days I’m building a meta-DB from the data that accumulates here and experimenting with various uses for it.
At first I just read it myself, then I shared it with the members of my AI study group and people in other departments at work, and got feedback from them. But running it among just a small circle meant limited perspectives, and the feedback loop didn’t really turn well. One comment I got recently — “I don’t get why this matters” — is what made me start adding background and a “so what” to every item, and it made me realize I need more feedback like that.
So I’ve decided to open it up publicly. I tried several versions, but in the end, a technology strategy and management perspective fit my interests and experience best, so this is a briefing organized around those two axes.
The daily edition delivers three items each morning with background; the weekly edition (beta) covers the week’s flow on Sundays; and the monthly edition (still in development) recaps the month at month-end.
If you’re interested, please sign up using the form below.
The image is an excerpt (the opening) from an actual issue sent out.
A heads-up in advance,
The daily edition goes out every day, so that’s roughly 30 emails a month (it may occasionally feel like spam). If it’s too much, you can unsubscribe anytime.
Also, since this is still a work in progress, a duplicate email might occasionally go out due to a testing error, and a short survey might go out from time to time as well.
I’d appreciate it if you could just keep those two things in mind.
#AI #AIStrategy #TechStrategy #AINewsletter #ProductLeadership #AIGovernance