Many of today’s pieces touched on accountability and ownership when AI plays a part in generation. Super important topic as we move past just the “create stuff” phase of agentic AI.
[article] The Sign-Off Layer Is Becoming the Real Engineering System. The idea here is that AI made generation cheaper, but not ownership. Excellent post.
[article] When Purpose Backfires. Don’t lead with purpose if your team is hampered by thwarted impact. When (unnecessary) bureaucracy gets in the way, your burn out those powered by purpose.
[blog] The Open Source Maturity Spectrum. Thoughtful analysis by Steve. Corporate open source is rarely altruistic, and I’m saying that as the person who leads Google’s open source programs office! For many of us, it’s a strategic lever.
[blog] Atlassian’s DESIGN.md is here: what we learned testing portable design context in practice. We shared this standard a bit ago, and Atlassian shares some of the good and bad of their experience.
[blog] Announcing the Agentic Resource Discovery specification. Another open source spec, this time from Google, GitHub, NVIDIA, and others. It’s about make it easier to discover and consume agentic resources.
[blog] Using Agents, Keeping Agency. Good one from Devin on the decision layer, and keeping agency in a world of agents. You still own the result, regardless of how the work was created.
[article] Tech debt, process gaps keep firms in AI ‘pilot purgatory,’ study finds. Lots of untapped AI value held back by internal operational weaknesses.
[article] Vibe coding can build your pipeline. It can’t explain it six months later. Another piece that reminds us that institutional memory is important. Adding that to persistent specs or other logs is going to be critical to understand past decision.
[blog] How to Track AI Agent Lineage and Manage State in Code Repositories. I like how Jason’s been thinking about this. Are git commits enough to understand what happened over time? Or are we missing other metadata about the session that generated the code, along with other lineage information?
[article] Agentic coding and persistent returns to expertise. AI can know how to build, but it needs us to know what to build. Some interesting research here from Anthropic.
[blog] The Hidden Powerhouse: Demystifying the BigQuery Storage Write API. Super dive into this more high performing API that makes it easier and faster to load data into your data platform.
Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:
Richard Seroter's Architecture Musings 
Leave a comment