In the AI era attention is the new gold and this is a good argument against reducing headcount. You need a pool of attention that is going to be available to run all the items in your roadmap very fast.

The shape is contracting Product engineering got complicated for good reasons. Product managers to translate intent. Designers to translate experience. Engineers to translate constraints. Each role existed to own a part of the feedback loop process because no one person could hold all at once. At least not at scale. That is the assumption AI is dissolving. The translation layers between roles were never the work — they were the cost of distributing it across people.

I just needs to reach the point of replicating human-quality. If we reach that barrier at any point, the trillions invested in AI are going to evaporate, and there will be an incredible market crash. If we don’t reach that barrier, then I can’t imagine a world in which capitalism makes sense. https://www.scarletink.com/ai-and-the-imminent-death-of-capitalism/

1. Single responsibility works best People tend to model agents on people: a product owner agent, an engineer agent, a QA agent. That fails for the same reason UserService failed. A role is not a scope. It is a bundle of unrelated jobs that share a title. Single-task agents hold up better. A backlog grooming agent. A release notes agent. A migration risk review agent. One job you can describe in a sentence, with a small tool surface and a clear output.

As a product manager, I have created several agents to delegate areas of my work. I have one for backlog grooming, one responsible for maintaining metrics and analytics and one for design reviews. The recent Claude Code upgrades have made this possible.

I have to admit… AI has made Jira finally bearable…

Having fun building a Backlog Grooming agent… Main observation is: The core of the agent is a git repo. The agent writes what it learns from existing Jiras in its context in the repo, and also logs every action it takes in the repo. Observing the commit history is observing the agent working.

Apparently, the sexiest setup right for vibe coding is Claude code integrated in Cursor’s terminal! Agent inception.

https://www.youtube.com/live/3e0zRE1Fvjw?si=8kg07OnK6UYyXiWq&t=967 In this demo, I showcase how MCP (Model Context Protocol design can significantly improve agent understanding and ultimately make the end user’s experience much easier and simpler. The demonstration highlights how simple concepts like “fetch my workspaces” fail and cause user friction, without the proper context by the MCP server.

AI is automation. As a result, the best way to offer an ai-powered service is via an API to support automated tasks.

AI has made traditional explicit data transformation obsolete

A list of key finding to illustrate the impact of AIP Guidelines in the areas of API consumption and API production: API Production: Increased Requirement Fulfillment: A controlled developer experiment showed that developers using API Improvement Proposals (AIPs) and an API linter achieved a 10% higher requirement fulfillment rate compared to those without any protocols. Feedback Utility: Among developers using the linter tool alongside AIPs, 75% found it beneficial, aiding significantly in refining their API designs.

• Be socially beneficial • Avoid creating or reinforcing unfair bias • Be built and tested for safety • Be accountable to people • Incorporate privacy design principles • Uphold high standards of scientific excellence • Be made available for uses that accord with these principles

AI turns right brain thinking into left brain.

https://www.youtube.com/watch?v=DuSS18NzjEw&t=96s&ab_channel=SAEAthens I was invited to the Life of Dev podcast and had a relaxed and insightful conversation on the API Designer role, software, AI and more

Last week, Reddit filed their S-1 to go public. At least 10% of their revenue - about $60m - comes from selling data to train Large Language Models. Data sales invert the business model of the Internet. Instead of Reddit building product experiences that create good advertising data to earn more on ads, Reddit will launch product experiences that produce more valuable data to feed to LLMs. The LLM vendors should pay more for better data.