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.
Back in the codebase
Mark Zuckerberg moved his desk into Meta’s AI lab and ships code ten hours a week. Sergey Brin came out of retirement to write code on Gemini. Down at manager altitude the same thing is happening: LeadDev’s Engineering Leadership Report found hands-on technical work among engineering managers jumped from 20 percent to 35 percent in a single year, and every leadership tier – tech leads, managers of engineers, managers of managers, CTOs – reports more time in the code than last year. Why? AI collapsed the cost of turning an idea into working software, so a leader can gather context, spike a proof of concept, and pressure-test assumptions between meetings. Spotify EM Emma Bostian and Bloomberg’s Paul Williams urge caution, though: the point of being back in the code is technical credibility, not output. The manager’s job is still vision, requirements, consensus, and an environment where the team grows. Pick up the non-critical-path work that unblocks people, and leave the critical path to the team. Player-coach is real now, but the “coach” part is still the job.
— LeadDev, 2m, #management, #engineering, #ai
Manage for the right reasons
Stephane Moreau sorts would-be engineering managers into three buckets: those pulled by a genuine interest in growing people, those pushed by boredom with their IC work, and – the new one – those fleeing what AI has made of engineering, the days now spent reviewing and supervising generated code instead of building. His warning is for the second and third groups: management is not the escape hatch it looks like. A survey of 617 engineering leaders found 22 percent at critical burnout and 47 percent of EMs thinking about going back to IC work within a year. Gallup puts natural management aptitude at about one person in ten, and Google’s Project Oxygen ranked technical skill dead last among the behaviors of its best managers. Moreau once hired a former CTO who quit after six months to go back to coding. If the appeal of management is mostly that it isn’t your current job, that’s a signal to fix the current job – or to accept that the discomfort is the industry shifting under everyone, managers included.
— Stephane Moreau, 6m, #management, #leadership, #careers
Tool users and tool builders
Marty Cagan expected AI tools to democratize product creation; he now says he was wrong, and the reason is a distinction borrowed from Benedict Evans: most people simply aren’t tool-builders. Anyone can feel a pain point, but far fewer can abstract from the pain to the general problem worth solving, and fewer still can design a solution that survives contact with sales, finance, compliance, legal, and the legacy systems it has to live inside. “You have to know a lot about sales to make good sales software, but being good at sales does not make you good at making sales software.” The customer knows the pain; knowing that a product should exist and how it should exist is a different person’s job. That job – the product role – doesn’t disappear because generating code got cheap. “AI changes the thresholds but not the problem.” Writing the code was never the hard part, and cheaper prototypes just move the bottleneck back to judgment – the same place it’s always been.
— Silicon Valley Product Group, 9m, #product, #ai, #strategy
Kinda bad software
John Battelle has watched the markets price in a “SaaSpocalypse” – AI eating every enterprise tool – and thinks it’s a misread. AI is just software, he says. Code, algorithms, data, compute. And today it’s software with real problems, like weak data persistence and painful enterprise integration, and those are the kinds of problems the industry has ground down before. Every prior apocalypse narrative (PCs, the internet, cloud, smartphones) ended with the software business refactored rather than replaced. “I can imagine no scenario where businesses and consumers abandon using software tools to get things done.” The leaders he talks to treat AI as an embedded capability that has to be woven into workflows, not a product that replaces them. The pushback in his comments is worth reading too: cheap autonomous code generation really is new, and the speed of development may be the real disruption. Either way, the practical response seems to be integrate, and don’t panic. Same as it ever was.
— John Battelle, 6m, #ai, #strategy, #software
Sunglasses for slop
Brian X. Chen diagnosed himself with slop-induced burnout – a feed full of suspiciously bulleted, emoji-studded posts – and spent a week testing Pangram, a $20-a-month detector that claims to catch AI-generated text 9,999 times out of 10,000. A University of Chicago study backs the near-perfect claim, and in Chen’s roughly 50 tests it never missed: it found the AI sentences he sandwiched into his own essays, cleared Dickens as fully human, and busted a LinkedIn poster who had written exactly one sentence of a long viral post (the poster confirmed this). Chen compared the feeling to the alien-revealing sunglasses in “They Live.” But images are another story: Pangram and a competitor both waved through widely debunked fakes, so photo verification still needs work. Meanwhile, model “watermarking” is coming to text. Anthropic said this week it will embed invisible watermarks in Claude’s output, with EU transparency rules pushing the rest of the industry the same way.
— The New York Times, 10m, #ai, #communication, #quality
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.
Weekly 1:1 prep assembles the briefing you’d otherwise burn Sunday night building: an overview of what each report worked on last week, a day-by-day breakdown, their current goals, any blockers they raised, and the pull requests they merged – delivered Monday at 11 AM, before the week’s conversations start. If you’re a player-coach now, spending more of your week in the code, this is how the coach half doesn’t slip. Walk into every 1:1 already holding the context, and spend the time on the person instead of the recap.
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.