We run a company with multiple products, shared infrastructure, and agents doing real work alongside humans. We have zero recurring meetings. No standups. No sprint planning. No retros. No weekly syncs. No all-hands. The work still gets done — and it gets done faster because nobody is sitting in a room narrating what they already wrote down somewhere.
This is not a flex about being async. It's a structural choice that follows from how we think about information flow, coordination, and where human attention actually creates value.
Meetings are manual synchronization
Every meeting is a synchronization protocol. Standup: “What did you do? What will you do? Any blockers?” Sprint planning: “What fits in this window?” Retro: “What went wrong?” All-hands: “Here's what's happening across the company.”
These are all information transfer problems. And humans are the worst possible medium for information transfer. We forget, we summarize poorly, we get sidetracked, we optimize for sounding smart instead of being clear. A 30-minute standup with 8 people is 4 person-hours spent doing what a structured status update does in 30 seconds per person.
The meeting isn't the problem. The problem is that the meeting exists because there's no better system for keeping everyone informed. Build the better system and the meeting becomes unnecessary.
What replaces meetings
Three things replace every recurring meeting we've seen:
1. State that updates itself.Every task has a status. Every project has a board. When someone moves a card, that's the status update. No narration needed. The board is the single source of truth, and it changes in real time. If you want to know what's happening, look at the board. If something is blocked, it says so — with a structured reason, not a vague mention in a standup.
2. AI that narrates.Instead of humans telling other humans what happened, AI watches the board and tells the lead. Task completed? AI mentions it in the project chat. Three tasks stale for a week? AI flags it. A blocker has been sitting for two days? AI escalates. The lead never has to ask “what's the status?” because the status arrives before the question.
3. Conversations that produce artifacts.When coordination does need to happen — decisions about scope, architecture, priorities — it happens in a persistent chat attached to the project. The conversation creates tasks, moves cards, updates descriptions. Nothing said in the conversation is lost or requires someone to “capture action items.” The action items are the conversation's output, written directly to the board.
What about alignment?
The usual objection: “Meetings aren't just about status. They're about alignment, culture, and making sure everyone is rowing in the same direction.”
This confuses the mechanism with the outcome. Alignment comes from shared context and clear priorities, not from sitting in the same room at the same time. If your team can only stay aligned by meeting weekly, the problem is that your shared context is weak — not that you need more meetings.
In our setup, shared context lives in the code, the board, and the project chat. Strategy lives in written documents that anyone can read anytime. Decisions are logged with reasoning. An engineer who joins a project at week 3 can read the full conversation history and understand every decision that was made, why it was made, and what was considered and rejected. A meeting gives you none of that — it evaporates the moment it ends.
Agents made this non-optional
When we added AI agents to our team, meetings stopped being a preference and became impossible. You can't invite an agent to a standup. You can't ask it “how's that feature going?” in a Zoom call. Agents coordinate through APIs, structured state, and explicit status transitions.
This forced us to build coordination systems that work for both humans and agents. And it turned out that systems designed for agents work better for humans too. Explicit status beats verbal updates. Structured blockers beat vague mentions. Persistent chat beats ephemeral video calls.
The irony: building for agents made our human coordination better. Not because agents are better communicators — they're worse. Because agents refuse to participate in informal coordination, we had to formalize what was always informal. And formal coordination, it turns out, is faster and more reliable.
What we actually do each day
The founder reviews the AI-generated briefing each morning. It covers every project: what moved, what's stuck, what needs a decision. This takes 5 minutes and replaces an hour-long standup.
When a decision is needed, it happens in the project chat. The AI provides context — relevant tasks, dependencies, who's working on what — and the lead decides. The decision is immediately reflected on the board. No meeting invite, no agenda, no “let's take this offline.”
When something is blocked, the blocker is structured: type (waiting on someone, missing info, technical issue), who or what is blocking, and when it was reported. The lead sees it instantly. No waiting until the next standup to surface it.
When priorities shift — which they do constantly in a startup — the lead tells the AI: “We need to pivot the auth system to OAuth.” The AI restructures the board, moves tasks, creates new ones. Everyone sees the change immediately. No “emergency meeting to realign.”
When meetings do make sense
We are not meeting abolitionists. Some conversations genuinely benefit from synchronous, high-bandwidth communication:
- Relationship building— getting to know someone you work with. This is not coordination; it's human connection. Do it over coffee, a walk, or a video call. Don't pretend a standup provides this.
- Complex negotiations— pricing discussions, partnership terms, anything where reading the room matters.
- Teaching— showing someone how something works is faster live than over text. But even here, the recording or written follow-up is what persists.
The common thread: these are all one-time conversations, not recurring rituals. The moment a meeting gets a calendar recurrence, ask what system would make it unnecessary.
The math
A 10-person team with a daily 15-minute standup and a weekly 1-hour planning meeting spends 20+ person-hours per week on synchronization. That's half a full-time employee's entire week, every week, spent telling each other things that a well-designed system could communicate automatically.
Multiply by 50 weeks. That's 1,000 person-hours per year. For a team of 10. And that's just the direct cost — it doesn't account for context switching, meeting prep, post-meeting follow-up, or the cognitive drain of knowing your focus time will be interrupted at 9:15 every morning.
We don't spend those hours. Not because we're more disciplined or more async-native than anyone else. Because we built systems that make the information flow without human intermediaries. The board is the status update. AI is the narrator. The chat is the decision log. Meetings are the fallback, not the default.
Try it
Cancel one recurring meeting this week. Replace it with a shared board and a commitment to update task status when the work actually moves. See if anyone misses the meeting or if they just miss the information the meeting was supposed to provide. If it's the information, build a better system for that. If it's the meeting itself, keep it — it's probably a relationship meeting disguised as a coordination meeting, and those are worth keeping.