Blurbs

I'm curious if anyone is building agent harnesses with idempotence via some mechanism where outbox and/or saga patterns are spun up on the fly and the critical agents execution goals are locked in until completion.

Those are big words, what does that mean Andre?

Imagine you specify your goal, and your agents come up with a plan. Instead of just spinning up a todo list they come up with real contracts for themselves where all obligations of all contracts are completed or none of it is considered complete.

I'm sure this is done in bespoke 'deterministic' agent flows, but I'm wondering if anyone is baking it into a general purpose harness.

I'm thinking out loud here, but I imagine this kind of thing could be composed with existing harnesses, skills and tools. That said, I'm wondering if a standard is emerging somewhere.

The naive or knee-jerk "can't teach an old dog new tricks" mentality might presume to close the door on 'expensive' experienced talent.

But capital W Wisdom is critical for maximum AI leverage. That Wisdom is gained over time. Knowing the right questions to ask and giving informed direction to people or agents yields better results that compound quicker.

I might ask the question of whether AI is accelerating or decelerating aging in.

Here too the naive "ai is as good as any junior and doesn't complain" mentality might presume to close the door on 'cheap' but inexperienced talent.

But once that Wisdom does actually age out, who in any business is left with institutional memory, domain knowledge, or real judgment?

I've been testing Gemma4 models running locally today. Believe the hype. They're fast and the smaller 4B model is doing a great job with NLP and NER tasks.

• e2b is super fast but overly concise
• e2b misses more secondary entities/typed mentions (lower recall).
• e4b gives richer extraction (more entities, more nuanced categories), but with more noise/normalization quirks - and takes about twice as long to process.

The results for the task were as good or better thang GPT5-mini. I'm not sure if OpenAI was having issues today, but performance was much faster than GPT5-mini over the network today.

Mid stage enshitification of GitHub and GitHub Copilot: Ads in pull requests. They walked it back, but the advertisements were an example of what I posted about yesterday: Abdication of responsibility to exercise good judgement.

To the Github Copilot team's credit they explicitly called out 'poor judgement' as the core issue and took some responsibility after the fact.

If you take their stated goal at face value the team was trying to surface 'helpful tips' to users. How the defined a 'tip' and where they chose to surface them both lacked appropriate judgement. To a point a friend recently commented, they met a specification (ads in pull requests) but not the requirements (helpful tips for users).

Zack Manson connected the dots to Doctorow for me with this quote on enshitification:

Cory Doctorow

I'm bullish on agentic AI. This turns out to be an unpopular opinion. A recent-ish survey shows only 10% of people in North America and Europe are more excited than concerned about AI.

Given the endless hype, the amount of FUD - fear, uncertainty, and doubt - in media reports, and the sheer volume of slop, I get why people have concerns.

That said, my optimism is 100% grounded in my personal experience over the last three plus years. One person's experience shouldn't convince you on its own, but if you're one of the roughly 50% that are more concerned than excited about AI, its worth asking what it would take to change your mind. Be a bit more curious about the possibilities. Try the contrarian viewpoint.

Should you build for the future or build for right now? I've done both. Plenty. In code and design. You eventually develop an intuition for which path to take. But today, all of those intuitions in your multi-objective optimization have to be completely re-assessed.

Have you ever shipped something without test coverage because the acute pain of not shipping now was greater than the an unknown future pain of a regression? Do you still not have standardized tokens in your design system because the acute pain of listening to developer refactoring complaints is greater than future reward of making their work easier?

Those intuitions about pain don't apply any more. You can have it both ways. You can ship right now, and have your agents continuously improve the testability of your system. You can codify design standards, because agents love that kind of clarity and don't complain about refactoring. In both cases, agents can continuously re-write documentation for future humans and agents alike.

Even your intuition of how hard it might be to build continuous agentic improvements into your day-to-day work is probably wrong. The cost of writing a skill, tool, or a full agent is approaching zero.

You just have to get started, and start building new intuitions.

Design is crucial. Period. Just enough design. Just in time.

What is just enough or just in time?

What problem are you solving? How irreversible are decisions you make? What's the impact of getting it wrong? Who's impacted? When do they need it?

Only you know how to find the answer to questions like these unique to your situation. And, don’t hate me for pointing out that how you get answers to questions like these is a process.

Just enough process.