You’re reading The Steady Beat, a weekly pulse of must-reads for anyone orchestrating teams, people, and agents across the modern digital workplace – whether you’re managing sprints, driving roadmaps, leading departments, or just making sure the right work gets done. Curated by the team at Steady.
Not the job
Charity Majors once had a boss who told her, “If your team is happy, you’re doing a good job.” In the third part of her back-and-forth with Dr. Cat Hicks, she explains why that’s wrong. Before Google’s Project Aristotle and DORA, tech celebrated firefighters and put up with brilliant jerks. Then the research showed that team dynamics drive delivery, and psychological safety became the thing to get right. Over time, the idea drifted. Psychological safety was supposed to support excellence and come paired with accountability. Now people cite layoffs, feedback, or an after-hours Slack message as threats to it. Majors puts it this way: “psychological safety isn’t important because it makes us feel safe, it’s important because it enables learning.” And learning is uncomfortable. Admitting you don’t know the answer in front of your peers is hard, which is why it builds trust. She says her generation of managers rejected command-and-control and overcorrected into “please be nice.” The best leaders, in her words, are “kind, caring humans and skilled business operators.” The team’s feelings matter, but they’re a consideration, not the target. “Making your team happy is not the job.”
— Charity Majors, 8m, #leadership, #management, #culture
Tool tired
“AI fatigue is hitting some development teams hard,” write thoughtbot’s Richard Newman and Michelle Taute. In their study of US healthcare technology leaders, 73% feel pressure to evaluate every new AI trend, and 62% say engineering fatigue is growing as teams manage AI-generated code alongside legacy systems. Another survey found 55.7% of tech workers report significant burnout, up from 44.7% a year earlier. The causes stack up: a stream of tools to evaluate with no clear goal, no consistent criteria for security and governance, more code to review, and decision fatigue from supervising agents. Then there’s the emotional part: “there can be significant grief in turning over this hard-won skill to AI.” They recommend practical fixes. Ask executives what they want, speed or security, and point to work that already delivers it. Evaluate AI vendors the way you’d evaluate any cloud provider. Run short pilots in a sandbox with fake data. Pick a review level on purpose, from reading every line to a “dark factory” that trusts the tests. (thoughtbot runs a “dim factory” – somewhere in between.) And “it’s also 100% fine to admit you don’t know a given tool yet.” Everyone has a cognitive load limit, including your best engineers.
— thoughtbot, 4m, #ai, #engineering, #burnout
Headcount math
A manager with about 100 engineers asks for 50 more. Should her director approve them? That depends on whether her current team works efficiently, and that’s hard to tell. “Measuring output and time spent is easy, while measuring efficiency and effectiveness is hard.” Anyone can count how long the work took. The real question is how long it should take: “Compared to how fast a good team would complete this task, how long did my team take?” And that changes with the kind of work. Routine production has a known scope, so you can estimate it. Invention doesn’t. David Anderson uses logo design to make the point. A generic one costs about $5 and takes a day, often with AI. A distinctive one costs thousands and takes as long as it takes, because “you literally don’t know what the good finished product looks like, because you’re searching for it.”
— Scarlet Ink, 16m, #engineering, #management, #metrics
Meter running
First came #tokenmaxxing, with companies pushing workers to use as much AI as possible. Then came the bill. In a recent study, only 11% of nearly 400 businesses could accurately forecast their AI spending, and The Wall Street Journal shows why it’s so hard. Price per token tells you little. Researchers from Stanford, Carnegie Mellon, UC Berkeley, and Microsoft Research ran models through more than 6,800 tasks, and in 32% of cases the cheaper model cost more. In one example, Google’s Gemini 3.1 Pro finished a task in 85 steps for about $1. The lighter Gemini 3 Flash took nearly 1,000 steps, ran up $14, and failed. “Price alone should not be used to infer which model is actually cheaper,” says researcher Lingjiao Chen. The same model varies from run to run, too. The Journal compares it to a lawn service that charges $20 an hour: two hours one week, eight the next, then five hours with half the lawn left uncut. AI behaves less like software and more like a worker on the clock, mistakes included.
— The Wall Street Journal, 6m, #ai, #costs, #strategy
Kill your gurus
George Mack’s essay makes one big claim: high agency may be the most important idea of the 21st century. He defines it as three skills working together: clear thinking, a bias to action, and disagreeability. Take one away and the whole thing stops working, like a tricycle missing a wheel. Low agency is the default, he argues, because brains evolved for scarcity and school was built for industrial work. The rest of the essay describes tools to make the the three skills work. For clear thinking, write a vague problem down or draw it. When you’re overthinking, flip the question. To bias for action, treat a decision as an experiment and break a big goal into levels, like a video game. For disagreeability, remember there’s no single right way to do anything and no infallible adult to defer to. “Kill your gurus,” says Mack.
— High Agency, 30m, #leadership, #agency, #productivity
Echo of the week
Echoes are AI agents in Steady that automatically gather and deliver work context to teams on a schedule – answering recurring questions about progress, capacity, and coordination so you stop burning hours assembling the same information manually.
Team capacity. Every plan assumes people show up. Time off lives in a handful of different systems, so the gaps usually surface the week they hit. Once a month, on a Monday at 11:00 AM, this Echo lists everyone’s absences for the next 30 days, grouped by week. You see the coverage gaps before you set a deadline on top of one.
The human-agent teamwork OS
Teams rely on two coordination loops to function: a big-picture loop connecting plans to progress, and a ground-level loop keeping teammates in sync.
Running those loops was already a marathon of meetings, chat threads, dashboards, and manual toil. Pile on flatter orgs, exponential output, and AI agents shipping 24/7 — the old way can’t keep up.
Steady removes the coordination bottleneck by running both loops for you. Working in the background, Steady distills updates and activity into targeted context for everyone on the team — human and agent alike. Full visibility, tight alignment, zero overhead.
The outcome: high-performing teams that deliver the right work, not just more of it.
Learn more at runsteady.com.