Ben Darfler

Craft can be applied to anything, including the humble chart.

Levels of Autonomy for AI Agents: operator, collaborator, consultant, approver, and observer.

"think of “conductor” versus “orchestrator” as two ends of a spectrum of AI-assisted development, with many hybrid workflows in between."

Thank you, Jade, for dropping the post I had in my head this week. I'm not on team 50 direct reports, but post ZIRP we are certainly pushing beyond the 6-8 that felt comfortable before.

Google’s Goals/Signals/Metrics (GSM) framework is a simple yet powerful tool to measure human systems. I’m surprised it remains buried in the conversation of Engineering Productivity and UX research and hasn’t expanded into areas like OKRs.

"Acceptance reduces resistance; resistance fuels persistence." The way to move above the line more often is to accept all the times when you can't.

"The next twenty years of observability are not going to be about prettier dashboards. They are going to be about what your trace structure says to a non-human reader at three in the morning, when you are asleep, and your agent is the only thing standing between your customers and an outage you will hear about on Monday."

5 functions of a manager: plan, organize, integrate, measure, and develop.

Some of the best teams I've been a part of have thrived because of the trust that was built through collaborative work. The more we dive into AI, the more I'm curious to see what collaborative practices emerge that we can pair with it.

You need to think about both: "Push communication ensures clarity and direction from leadership, aligning teams with the overarching goals. … Conversely, pull communication empowers team members to proactively seek information, driving bottom-up innovation."

- Principle 1: Citations are the Currency of Trust - Principle 2: Designing for the Editor State - Principle 3: The Power of ‘I Don’t Know’

Mentorship programs that develop theory-building skills

I appreciated the 7 practical suggestions that ground what can be a fluffy topic: - Demonstrate Leadership Commitment and Role Modeling - Encourage and Reward Collaboration - Promote Psychological Safety - Implement Blameless Postmortems - Strive for Information Transparency and Clarity - Celebrate Innovation, experimentation - Foster Continuous Learning

I appreciated the set of patterns (very on brand for thoughtworks): • Knowledge Priming • Design-First Collaboration • Context Anchoring • Encoding Team Standards • Feedback Flywheel

I appreciated the terms "attended" and "unattended" parallelism as well as this observation: "The engineer’s job … [is] splitting into two distinct halves … building the environment … [and] managing the work"

We are shifting from generating more code to verifying more code.

Nice framing for giving agents more programmatic feedback and needing less human feedback.