You’re reading The Steady Beat, a weekly pulse of must-reads for anyone orchestrating teams, people, and work 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.
Faultlines
AI is creating a new fault line in organizations, and it’s not between humans and machines. It’s between the people building with AI and the people betting on it. Gregor Ojstersek identifies three tensions tearing at team cohesion: executives expecting plug-and-play magic, PMs and engineers talking past each other on timelines, and speed-obsessed AI adopters clashing with methodical traditionalists on the same engineering team. The executive gap is the most dangerous; leaders driven by FOMO demanding “AI transformation” without defining which business problem they’re actually solving, while engineers who’ve survived enough hype cycles respond with reflexive skepticism. Meanwhile, PMs hear “AI can do that” and translate it into aggressive timelines that ignore integration complexity, and engineers who’ve been burned by over-commitment dig in harder. The prescription is perspective. Before you dismiss your PM’s optimism as naivety, understand they’re getting squeezed by the same unrealistic expectations from above. Before you write off engineering’s caution as resistance, remember they’re the ones holding the bag when the “quick AI win” becomes a six-month slog. The organizations navigating this well are running small experiments, defining problems before tools, and building the psychological safety required for honest conversations about what AI can actually do today versus what the pitch deck promises.
— Engineering Leadership, 6m, #ai, #leadership, #coordination
The Company Is a Brain
What if your org chart is already obsolete? Not because of bad management, but because the entire premise of organizing humans into departments is about to be replaced by something closer to a nervous system? Scott Belsky’s vision of the “cognico” is provocative and worth sitting with: future companies built around an AI nucleus of compute and data, with specialized function nodes replacing traditional departments. In this model, humans don’t disappear, but they radically change shape. Belsky identifies three surviving roles: stewards who ensure model quality and data integrity, orchestrators who design workflows across AI systems and APIs, and leaders who provide the heart — vision, culture, and the irreducibly human judgment that no model can replicate. The real challenge for today’s managers is tackling the intermediate step. Customer-facing functions are already collapsing: support responses become marketing moments, social complaints become sales opportunities, and the old departmental walls look increasingly arbitrary. Meanwhile, competitive advantage is quietly shifting from who has the best people to who has the best proprietary data and the smartest rules governing how AI uses it. You don’t need to reorganize your company tomorrow, but you should probably stop assuming that your current structure will survive contact with what’s coming.
— Implications, 13m, #ai, #transformation, #strategy
Context on Autopilot
Steady’s latest update doubles down on a simple premise: the best coordination happens when you stop hunting for context and let it come to you. The headline features are Insights integration in Echoes and RSS feed support — two capabilities that sound incremental but fundamentally change how teams stay informed. Echoes can now pull real-time analytics directly into scheduled deliveries: participation rates, team performance leaderboards, comprehensive metric snapshots — all without anyone manually checking dashboards or assembling reports. Need to know which teams are falling behind on engagement? Set it up once and it just shows up. The RSS addition is equally practical — point an Echo at industry blogs or competitor sites, and get AI-filtered summaries on a schedule instead of maintaining yet another tab graveyard. Under the hood, Insights queries are running up to 80% faster with async card loading smoothing out multi-report views. The throughline across all of these updates is the same: automate the gathering so humans can focus on the deciding. If your team is still spending Monday mornings pulling together status reports and scanning feeds, that’s coordination tax you don’t need to pay anymore.
— Steady Blog, 4m, #productivity, #coordination, #ai
Engineer the Harness
Mitchell Hashimoto – the creator of Vagrant, Terraform, and Ghostty – isn’t here to sell you on AI hype. He’s here to tell you that chatbots are mostly useless for real coding work, and that the actual productivity gains require a fundamentally different approach. His journey from AI skeptic to power user follows a surprisingly predictable arc: abandon chat interfaces, switch to agents that can read files and run code autonomously, then systematically engineer the scaffolding that makes those agents reliable. The killer insight is what he calls “harness engineering”: creating AGENTS.md files and custom verification tools that prevent recurring mistakes, essentially teaching your AI collaborator through documentation rather than conversation. Break work into discrete, verifiable tasks. Give agents a way to check their own output. Run them in the background on research and low-risk implementation while you focus on the work that actually requires your brain. Hashimoto’s honest about the limits too: knowing what agents can’t do saves as much time as knowing what they can. He’s currently running agents 10-20% of the time in the background – not the “AI writes everything” fantasy, but a pragmatic multiplier that compounds over weeks. The real takeaway for managers? AI adoption isn’t a light switch. It’s a skill that develops through deliberate practice, workflow redesign, and the humility to acknowledge that your first attempts will be terrible.
— Mitchell Hashimoto, 13m, #ai, #engineering, #productivity
Hold It Lightly
Every team has that one technical decision from month two that everyone knows is wrong but nobody wants to revisit because “we already committed.” Andrew Bosworth, Meta’s CTO, makes a seemingly simple case: hold your goals firm and your implementation loose. Your earliest decisions are made with the least information, which makes them the worst candidates for being treated as sacred. Yet that’s exactly what most teams do. They fall in love with the architecture, the framework choice, the sprint plan – and then spend months defending sunk costs instead of adapting to what they’ve learned. Bosworth’s distinction between the why (stable) and the how (adaptive) is the kind of thing that sounds obvious until you watch a planning meeting devolve into a turf war over implementation details nobody will remember in six months. Strong teams treat early decisions as hypotheses, not commitments. They course-correct without existential crisis. This flexibility isn’t indecision, it’s respect for the fact that you’re smarter now than you were when you started. The next time someone says “but we already decided,” ask them: decided what, and decided why?
— Boz, 3m, #leadership, #strategy, #engineering
Toxic on Purpose
The behaviors most leadership content tells you to avoid might be exactly what your team needs. Gaurav Jain makes a provocative case that directness, high standards, tight deadlines, and emotional boundaries (the stuff that gets you labeled “toxic” on Glassdoor) are actually the hallmarks of leaders whose teams perform best. The argument isn’t a license to be cruel; it’s a challenge to the cult of comfort that’s invaded management culture. When you soften feedback so much that the message disappears, you’re not being kind – you’re being cowardly. When you set standards everyone can comfortably meet, you’re not being reasonable – you’re telling your team you don’t believe they’re capable of more. The McKinsey data backs this up: clarity of expectations is the single biggest driver of engagement, not warmth or approachability. The sharpest distinction Jain draws is between genuinely toxic behavior – abuse, fear-mongering, belittling – and high-standards leadership with care underneath it. Protecting your time isn’t selfish; it models the boundaries you want your team to set. Being strategically unavailable isn’t neglect; it’s forcing your team to develop the problem-solving muscles they’ll never build if you’re always there to catch them. Teams don’t want comfortable bosses. They want bosses who believe they can do hard things.
— The Good Boss, 7m, #leadership, #management, #productivity
The Exhaustion Paradox
Here’s the reality of AI-assisted work: you’re faster at every individual task and more exhausted than ever. Siddhant Khare names the thing nobody in leadership wants to acknowledge: AI fatigue is real, it’s widespread, and it’s the predictable result of tools that reduce the cost of production while increasing the cost of coordination, review, and decision-making. The pattern is insidious. AI generates code in seconds, so expectations expand to fill the gap. You’re no longer a creator; you’re a quality inspector, reviewing output you didn’t write for bugs you can’t predict because the same prompt produces different results every time. The nondeterminism alone would be maddening, but stack on the FOMO of a tool ecosystem that reinvents itself monthly and the perfectionism trap of outputs that are almost right (worse than completely wrong because you can’t stop tweaking them) and you’ve got a recipe for burnout that looks like productivity from the outside. Khare’s most actionable insights: limit AI refinement to three attempts, then write it yourself. Time-box AI sessions to 30 minutes. Dedicate mornings to independent thinking before the tools get involved. The real skill of the AI era is knowing when to stop. Your team’s cognitive resources are finite, and no amount of tooling changes that math.
— Siddhant Khare, 16m, #ai, #engineering, #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.
Feed Summarizer — Point this Echo at any RSS feeds—company blogs, competitor sites, industry publications, release notes—and get a concise weekly digest of everything new. No more tab-hoarding or falling behind on what matters. It just shows up, summarized and ready to scan.
Stop Drowning in “Work About Work”
Your team loses 21 hours per person, per week slogging through status meetings and hunting for context across chats, docs, and dashboards.
Steady’s AI agents eliminate this coordination tax by continuously delivering personalized guidance on what’s happening, what’s next, and what needs attention across the whole team.
Join the thousands of teams staying in sync, avoiding burnout, and moving 3X faster with Steady.
Learn more at runsteady.com.
