Your calendar treats a status update and a make-or-break strategy call as the same object: a 30-minute video block. It shouldn't. A clear, look-it-up question and a tangled, no-right-answer problem are not the same kind of work — and pouring both into the same recurring meeting is why so many of your calls stall. The fix isn't fewer meetings. It's matching the meeting to the problem, and the cleanest way to do that is the Cynefin framework for meetings.
The numbers say teams almost never do this. 52.9% of meetings are recurring — the same room, the same cadence, no matter what's actually on the table. Worse, 77% of meetings end by scheduling another meeting, and Atlassian found meetings are ineffective at accomplishing their goal 72% of the time. That's not an overload problem. It's a sorting problem.
Below: what the Cynefin framework actually is, why distributed teams default to one mode for everything, and how to route each of the four problem types — clear, complicated, complex, and chaotic — to the right response. Plus where AI helps, and where it can't.
What the Cynefin framework for meetings actually is
Cynefin (Welsh for "habitat," pronounced kuh-NEV-in) is a sense-making framework created by Dave Snowden in 1999 and popularized in a 2007 Harvard Business Review article with Mary Boone. It sorts problems into five domains based on the relationship between cause and effect: Clear (once called Simple or Obvious), Complicated, Complex, Chaotic, and a center called Confusion or Disorder.
The split that matters most is "ordered" versus "unordered." Per Wikipedia's reference, the Clear and Complicated domains are ordered — "cause and effect are known or can be discovered." The Complex and Chaotic domains are unordered — "cause and effect can be deduced only with hindsight or not at all." Each domain has its own action sequence, and using the wrong one is how smart teams get stuck.
Snowden's core point is the one your calendar ignores: problem solving is not a one-size-fits-all process. A meeting is just a container. The question is which kind of problem you're putting in it — and most teams never ask.
Why remote teams confuse complex and complicated problems
The recurring 30-minute call is the path of least resistance. It's already on the calendar, so every problem gets poured into it — the clear status update, the complicated trade-off, and the genuinely complex decision all get the same container, the same time box, and the same passive format. That mismatch is expensive. McKinsey estimates inefficient decision-making costs a typical Fortune 500 company 530,000 days of managers' time a year, about $250 million in wages, with 68% of middle managers saying their decision-making time is used inefficiently.
Remote work amplifies the problem, because the default reflex is to "hop on a call" rather than choose a format. Microsoft found 57% of meetings are now ad hoc calls with no invite, employees are interrupted every two minutes, and 48% of workers say their work feels chaotic and fragmented. When everything is a quick sync, nothing gets the format it actually needs.
The Cynefin framework breaks the reflex by forcing one question before you send the invite: what kind of problem is this? Answer that, and the right response — async note, expert review, live working session, or emergency call — picks itself. Here's how each of the four domains maps onto a real meeting decision.
Clear problems: don't hold a meeting at all
A clear problem has a known, repeatable answer. The action sequence is sense–categorize–respond: you look at it, match it to a known category, and apply the best practice. Status updates, FYIs, routine approvals, and obvious reversible decisions all live here.
In Cynefin terms, the Clear domain is the realm of "known knowns," and the rule for it is blunt: if the answer is knowable by reading something, it's a message, not a meeting. GitLab's all-remote handbook is explicit about this — it keeps status updates and FYIs out of synchronous time entirely, reserving live time for problems that genuinely need it. A recurring status call is the single most common way teams waste the clear domain: they gather six people to recite information that's already sitting in a project tool.
Convert these to async by default. If you're still defending a standing status meeting, you're probably running a status meeting where a working session belongs — or, more often, holding a meeting for work that should never have been one.
Complicated problems: bring the experts, async-first
A complicated problem has a right answer, but you need expertise to find it. The sequence is sense–analyze–respond, and the goal is good practice rather than a single best practice — there may be several valid solutions. Choosing a database, pricing a new tier, or scoping an integration all sit here. This is the heart of the complex vs complicated problems distinction: complicated is hard but knowable; complex is genuinely unpredictable.
The mistake teams make in this domain is convening a big live meeting to "discuss" something that's really an analysis task. Most complicated problems are better served by an expert doing the analysis async — writing up the options, the trade-offs, and a recommendation — and circulating it for asynchronous review. You only need a live call when the experts disagree and the disagreement needs to be worked out in real time.
So the test for a complicated problem is: can one or two qualified people analyze this and propose an answer the rest can react to? If yes, it's a document and a team decision thread, not a standing meeting. Save everyone's calendar for the domain that actually requires the room.
Complex problems: probe, sense, and respond in a live session
A complex problem has no right answer you can know in advance. Cause and effect only become clear in hindsight, so the sequence is probe–sense–respond: you run small, safe-to-fail experiments, watch what happens, and amplify what works. As the canonical reference puts it, in this domain "instructive patterns emerge if the leader conducts experiments." This is emergent practice — you create the answer rather than retrieve it.
This is the meeting worth protecting. Strategy pivots, ambiguous product bets, cross-functional problems with no obvious owner — these need people in the same space, thinking out loud, building on each other. The work is divergent (generate many possibilities), then convergent (cluster, test, narrow). A virtual brainstorming session or a remote design sprint is the complex domain done right.
It's also where a passive video grid fails hardest, because probe–sense–respond needs a shared surface, not just talking heads. A tool that combines video, a live collaborative canvas, and a contextual AI that follows both — the model Coommit is built on — turns a complex meeting into an actual working session: ideas go on the canvas, get clustered and challenged in real time, and the thinking is captured as it emerges. The complex domain is the reason synchronous meetings still exist. Spend your live hours here.
Chaotic problems: act first, then meet
A chaotic problem has no discernible relationship between cause and effect — an outage, a security incident, a PR fire. The sequence is act–sense–respond: you act immediately to establish order, sense where stability returns, then respond. Cynefin's guidance for this domain is vivid: "a leader's immediate job is not to discover patterns but to staunch the bleeding."
The failure mode in chaos is treating it like a complex problem and brainstorming options while the building burns. You don't run a divergent workshop during an outage. You put one decision-maker in charge, open a fast synchronous room, take decisive action to stop the damage, and only later — once it's stable — analyze what happened. That analysis belongs in a remote postmortem, not in the heat of the incident.
So the Chaotic domain is the one place where "hop on a call right now" is the correct instinct. The discipline is knowing it's the exception, not the default — and not letting the urgency of a true crisis become the excuse for treating every problem as one.
The fifth domain: when you don't know which one you're in
The center of the model is Confusion (Disorder): you don't yet know which domain a problem belongs to, so you fall back on whatever mode you're most comfortable with. For most remote teams, that comfort mode is the status meeting — which is exactly why ordered-thinking gets misapplied to unordered problems. It's no surprise teams waste 25% of their time just searching for answers when the problem was never sorted in the first place.
The way out of Confusion is to break the problem into pieces and route each one. The reversible parts go async (clear). The parts needing expertise get an analysis (complicated). The genuinely uncertain core gets a live working session (complex). Suddenly a vague "let's all meet about this" becomes one short doc, one expert review, and one focused 45-minute session — instead of three weeks of circular calls. This is the same instinct behind work triage: sort first, then act.
Where AI fits — and where it doesn't
AI maps neatly onto Cynefin, which is exactly why it's so easy to misuse. In the clear and complicated domains, today's AI is excellent: it categorizes, summarizes, retrieves, and analyzes — the ordered work where a knowable answer exists. Microsoft reports 83% of leaders believe AI will let employees take on more complex, strategic work earlier in their careers, precisely because it absorbs so much of the ordered load.
The complex domain is where the limit lives. AI can create what some practitioners call "AI bubbles" — pockets where its pattern-matching temporarily makes a complex problem behave like a complicated one you can analyze. But genuine complexity still requires human probing, and treating an unordered problem as ordered is the original Cynefin mistake. (Snowden himself revised parts of the model in 2024, a useful reminder that it's a living framework, not a 2007 fossil.) The right role for AI here is as a contextual partner inside the probe–sense–respond loop — one that sees both the discussion and the canvas, surfaces patterns, and keeps the thread — not an oracle handing you the answer. That's the difference between AI that summarizes a meeting and AI that helps you think through one.
Match the problem, and the meeting takes care of itself
The number of meetings on your calendar was never the real problem. The mismatch is. When a clear status update and a complex strategy call get the same 30-minute box, the status call wastes everyone's time and the strategy call never gets the depth it needs. The Cynefin framework for meetings fixes both ends at once: sort the problem by its domain first, and the format — async note, expert review, live working session, or emergency room — chooses itself.
Do this for one week and watch what happens. The "could've been an email" meetings vanish, and the handful of genuinely complex problems finally get a focused, well-run session — the kind of meeting worth keeping. That's the whole point of a tool like Coommit: make the few meetings that survive the sort actually produce the answer. Match the problem to the response, and your team stops paying for the wrong meeting and starts winning the right one.