- Pointer
- Posts
- Issue #719
Issue #719
Essential Reading For Engineering Leaders
Tuesday 26th May’s issue is presented by Unblocked
Unblocked turns code, docs, tickets, and conversations into actionable context, so engineers move faster and agents stay on track.
— Michael Lopp
tl;dr: A senior leader's real job is figuring out what people actually need from you - "The Ask" - across three types of meetings: the easy ones (someone on your team wants a promotion), the cross-team ones (another team needs to build with you but doesn't know how), and the long-game ones (two leaders meeting regularly with no obvious purpose until, years later, the reason reveals itself). The deeper point is that much of senior leadership runs on instinct and pattern recognition built from years of experience, not clean data and well-defined agendas.
Leadership Management
— Ajey Gore
tl;dr: “For thirty years we were glorified translators - business asked why, product defined what, engineering translated to how. AI just ate the translation step. The anatomy of the team that's left looks nothing like the one you have today.”
Leadership OrgDesign AI
— Dennis Pilarinos
tl;dr: For agents to work at scale, they need to deeply understand how your team works. Rules, skills, and separate MCPs give access to information, but not understanding. Read how a context engine gives agents exactly what they need for the task at hand.
Promoted by Unblocked
AI DevEx
— Wes Kao
tl;dr: Ask yourself: (1) If I were signing off with my name, would I make this better? (2) Is there sloppiness that might reflect poorly on me or my company? (3) Am I passing the burden to others to fix my errors? (4) Does the work I’m shipping represent my ability?
Management CareerGrowth
"Teach for the future; you have to live in it".
— Andi Roberts
tl;dr: “How do you stay resilient, motivated, and mentally healthy in a difficult job with poor management, shift work, or constant stress? This guide explores practical, science-informed ways to protect your well-being while navigating a demanding current role.”
Wellbeing
— Santosh Kumar Radha
tl;dr: If your team has chained Claude Code to write, then to review, you already know the pattern. Harness orchestration is the architectural discipline behind it, a first look at how Claude Code, Codex, and Gemini compose as primitives. Read the breakdown — written for engineering leaders evaluating the agent layer.
Promoted by AgentField.ai
Architecture AI
— Marc Brooker
tl;dr: “I’ve been spending a lot of time thinking about the shape of the capabilities of coding agents. What they’re good at now, what they’re going to be good at. What they’re bad at now, how much of that is inherent and how much is transient. This is worth thinking about, because it’s the most important question shaping the future of software, and of software engineering. I don’t pretend to have an answer, but am coming to a conclusion that may be deeply counter-intuitive.”
SystemDesign AI
tl;dr: This article walks through how the Pinterest engineering team redesigned their user‑sequence platform to make sequences cheaper to run, faster to extend, and easier to debug, while still supporting demanding production use cases.
Scale Infrastructure ML
— Simon Willison
tl;dr: “I’ve started a new project to collect and document Agentic Engineering Patterns—coding practices and patterns to help get the best results out of this new era of coding agent development we find ourselves entering.”
DevEx AI Guide
Editorial Note
A roundup of other interesting articles and events on my radar:
Mitchell Hashimoto believes some engineering leaders are under AI Psychosis over-relying on agents to fix production bugs quickly without repercussion.
Isaac Van Doren is launching an independent conference on software reliability called Software Should Work (July 16-17, Columbia, Missouri). Speakers include Andrew Kelley (Zig), Richard Feldman (Roc), Richard Hipp (SQLite), Carson Gross (HTMX), and Marianne Bellotti.
Jose, who received his first Pointer issue back in 2017, and Dan are working on OpenProse, an open-source declarative language for AI agents. The thesis: rather than scripting what an agent does step by step, you declare the desired outcome and let the framework handle the rest. They're looking for engineer feedback.
The market for Forward Deployed Engineers - a hybrid role that mixes of software, sales, and platform engineering skills - is heating up again, largely driven by big tech.
Simon Willison published “The last six months in LLMs in five minutes”, with this standard ask of various LLMs to draw a pelican riding a bike. The interesting bit are the pelican on bike drawings.
Research on how LLMs can subtly distort our written language by producing larger semantic shifts, even when prompted to make grammatical edits, compared to human edits.
Most Popular From Last Issue
AI Productivity Debate - Justin Reock
Notable Links
AX: Google's OS distributed agent runtime.
Files: Simple app for your .md files.
FreeDomain: Free domain for everyone.
Plugins: Claude plugins for knowledge workers.
Understand-Anything: Turn code into a knowledge graph.
How did you like this issue of Pointer?1 = Didn't enjoy it all // 5 = Really enjoyed it |