As we use AI workflows more and more the bottleneck has become checking, reviewing, decision making, researching, contemplating which is good in the traditional sense of productivity. But it feels like the real question is if its even worth doing so much (managing 10’s of agents) for so little in return; like is most of this productivity even real? There is a sense that the output increased 10x but real gains are maybe close to 2x.

The decision to be deliberate with what we do seems more present today, and just because we can try 20 things doesn’t mean we should. Although its feels like I should know better.

Know of at least 3 people who built many different product prototypes quickly but didn’t get anywhere further on those projects. Maybe its best to build something you will use yourself.

Loops and other agentic workflow automation is great for a small sample of very well-defined goals and task but longer running agentic workflows always tend to end up with solutions so far off from what I have imagined due to nature of delegating the decision making to the agent.

Reliable local AI is still not easily accessible for most consumers and seems to be getting worse with the hardware prices today. This is further made worse with openclaw being so unreliable, both consumers and small businesses have been burned by this experience.

Job market is definitely tough right now. Many people are hanging onto their jobs right now, who in better times will jump every 1-2 years. Layoffs are clearly reaching within friendly circles and impact can be directly felt.

Posting on X feels like a dead end for the most part. Still good for reading opinions from a few thought leaders. Can’t believe I renewed for 500 bucks.

Watched The Odyssey on IMAX, pretty good overall and I like it but storywritters and philosophers will likely differ (which I understand). Can’t wait for Dune III.

Fun read recently On Bullshit by Harry G. Frankfurt