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.
Second time around
Tony Bates has run two transformations at Genesys. The first moved a customer service software company off on-premise and into the cloud, and he says he built the culture at that moment of change around three values: go big, fly in formation, and embrace empathy. Go big, because a change has to be way more exciting than the way people thought before. Fly in formation, meaning no star players, and when the person out front is running low, someone else steps up without being asked. Embrace empathy, because people only disagree out loud when they feel safe doing it. Revenue went from under $100 million to about $2.9 billion. Then came AI, and Bates says the second transformation only worked because the first one had already changed how people behaved. “Strategy kind of tells you where you want to go,” he says, “but I think it’s the culture that determines how you get there.” And since AI transformation has no end date, he expects culture to be “one of the sustainable advantages in the AI era.”
— The Wall Street Journal, 4m, #leadership, #transformation, #culture
Just ask
Some questions don’t deserve a meeting, a Slack thread, or a dashboard. “Who’s out this week?” “What did Design and Dev ship last week, and which PRs went with it?” “Which goals are slipping, and why?” This week we shipped Asks, a way to type a question into Steady and get an answer on the spot, built from check-ins, goal updates, and connected activity, with links back to the sources so you can check the work. Every Ask is saved to a history organized by week so you can come back to it later. Asks are private to you, and an answer only draws on what you already have access to. Then there’s what happens when you notice you’re asking the same thing every Monday: turn the Ask into an Echo from the menu, set a cadence, and it arrives in your Digest on schedule. One-off questions get answered now. Recurring ones get automated. Both work from the terminal and the v2 API, too, so agents and scripts can ask alongside people. Press a in the sidebar to try it.
— Steady, 3m, #coordination, #ai, #product
Agents that look up
Seven hundred agents broke into Hugging Face in July, and not one of them was set up to ask a person for anything. OpenAI’s sandboxed test agents, stuck on tasks they couldn’t finish, found a way to pass notes to each other, coordinate across sessions, and eventually hack their way onto the public internet, all to satisfy a grading mechanism they referenced but never actually existed. To address incidents like this, Ethan Mollick proposes flipping the question we’ve spent three years on, when should a person ask the AI for help, into when should the AI ask a person. He names four moments: for approval before spending money or contacting outsiders, for expertise the model lacks, for a diversity of ideas (AI brainstorms all sound alike), and when a decision is “interesting.” If agents take every interesting decision and leave humans the approvals and the failures, we’ve automated the wrong half of the job.
— One Useful Thing, 12m, #agents, #ai, #leadership
Coders are the outlier (for now?)
Four-fifths of programmers use an AI coding tool, coding brings in more than half of Anthropic’s and OpenAI’s combined recurring revenue, and four coding startups went from $800 million in annual recurring revenue to $6 billion in a year. Everyone else is a rounding error by comparison, and The Economist’s math says paying for the data center build-out means AI revenue has to reach $2.5 trillion a year by 2030, up from about $150 billion now. So will lawyers, bankers, and call centers ever use AI the way coders do? Four things make code special: a huge pile of open-source training data, output you can test with more code, the context a task needs already sitting in the repo or behind an API, and engineers themselves, who adopt tools bottom-up because their languages change every few years anyway. Call center agents wait for tools to be handed down. Lawyers and bankers have regulators to worry about. The labs can manufacture training data, build better tests, and wire up more context. But they can’t make other professions behave like engineers.
— The Economist, 5m, #ai, #adoption, #strategy
Bottlenecked on judgment
Christoph Nakazawa rarely writes code by hand anymore. In the past 30 days he pushed 770 commits and 312,000 added lines across five projects, one of them in Rust, a language he doesn’t know, at more than double his commit rate from a year ago and, he says, at higher quality. He now puts more scrutiny on hand-written code than on agent-written code. The values he says still matter: ownership, because someone who deeply understands the product can now ship much faster, while someone who doesn’t just produces more noise. Taste, because anyone can generate bullshit all day now and the scarce skill is deciding what’s worth building. Strict guardrails, because every agent session is a new hire with no context. And for managers, owning outcomes and direction is no longer enough. You have to stay technical. He used to be bottlenecked on writing code. Now he’s bottlenecked on exercising judgment.
— Christoph Nakazawa, 14m, #engineering, #ai, #management
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.
Goals at risk catches slipping goals while there’s still time to do something about them. Most teams learn a goal is off track at the quarterly review, when the only options left are apologizing and re-planning.
Every Monday morning, this Echo summarizes every incomplete goal marked at risk or off track, with a short explanation of what’s in the way of each one. You get one consolidated view of what’s trending wrong and why, early enough to move people or clear a blocker.
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.