The behind-the-scenes notes sharing system cannot be another high-fidelity written-word solution. I need a different approach because what I'm doing is just sharing important things as they happen. And it's also a really cool way for me to extend my ‘voice in written’ content system.
What the experiment is going to try to create is shown a little bit in the diagram below. Which is a scan of my paper sketch.
The reason I need a different flow. On a working day, I sometimes have two to three moments where I'm going, "Oh, this would be a useful share or a useful record for me personally to have." What's happening at the minute is that those moments are taking about half an hour to 45 minutes of typing at the end of the day.
Which is not, firstly, efficient or, secondly, very in line with the intent of becoming an agentic enterprise.
What I am going to do is create an agentic flow whereby these voice notes can be recorded and shared very quickly. The human judgement is still - is this a helpful share / note.
These are then pushed through a series of agentic loops and returned as an end-of-day summary, either an aggregated note or a series of notes. Again, that'll be another agentic decision in the process.
In my mind, what'll be very cool here is looking at the friction that that removes, but obviously the rule of thumb needs to be that the notes that are recorded are actually genuinely useful for someone following the experiment to get some insight on what is possible. Or for me to have a record of evolution and compounding improvement.
When is AI Content Not AI content?
The cover image in this note raises an interesting question. Does the fact that I put a scan of my original sketch and got AI to tidy it up. Land the result with the bad ‘Generated with AI’ label / hidden watermark?Or is it still, as I believe, my work and does not warrant the ‘generated with AI’ label. Who put Google in charge again?
That question itself shows you how quickly all of our systems of work are changing.