The AI OODA Loop Decider skill for Claude

The problem it solves: paralysis at the moment when movement counts. The competitor lowers their price and you schedule a meeting. Too many options land on the table and nobody cuts through. Or the opposite, just as costly: a decision gets made at speed based on a picture of the situation that has long stopped being true. In both cases it forces the same rhythm. First observe without judgement, then test your own lens, then a maximum of three options, then one concrete step today, and an agreed moment to look again.

It is written for anyone who has to decide under uncertainty while the situation keeps moving: business owners, marketers, sales teams watching a deal tip over, teams that have to contain a crisis. The file itself also states what it is not for: no long term strategy, a strategic thinking framework belongs there, and no root cause analysis, the 5 Whys method is for that. It is part of the skill library NL that we make freely available from the AI and automation service, with no account and no sales email afterwards.

And one thing upfront, because the file itself hammers on it: the loop is a thinking rhythm, not a magic formula. Nowhere does it promise that a faster cycle leads to winning. What it does do is stop you standing still while reality keeps moving, and stop you sprinting hard in the wrong direction because nobody tested the assumptions.

Boyd's most famous tactical idea follows from that: getting inside the opponent's OODA loop. Whoever decides faster than the opponent breaks their picture of the situation and forces them to react to something you have already moved past. You can see that again in the last line of FIG.01: while you run your loop, the competitor runs theirs, and the difference in rhythm decides who reacts to whom.

But the part of Boyd's work that gets left out most often is exactly the part this skill puts at the centre. Boyd himself called Orient the most important part of the loop, because that is where observation turns into understanding. He described that process in his 1976 essay Destruction and Creation: breaking down old mental models to build new ones. Speed without orientation is only getting to the wrong place faster. Tempo only counts once the direction is right.

To be fair, and the file says this itself: Boyd deliberately never published a book about the model. It lives on in essays and briefings, with Patterns of Conflict and The Essence of Winning and Losing as the best known, and was later worked out further by biographers and strategists such as Robert Coram, Chet Richards and Frans Osinga. The popularity of the loop in business literature is greater than the amount of empirical research behind it. Treat it as a tool, not as proof, because that is how it is used here too.

After that the file instructs, phase by phase, which questions Claude asks. At Observe: what facts are on the table, what has changed with competitors or clients, what measurable data exists, what unconfirmed signals are coming in, what is standing still and what is moving. With the fixed rule to separate hard from soft: a number and a rumour never get the same weight. At Orient, the longest part of the file: which mental model are you applying, which untested assumptions sit underneath it, what would your biggest competitor do with the same data, which pattern do you recognise from earlier situations and does that comparison really hold, what new synthesis emerges if you combine the data differently.

At Decide the file explicitly warns against analysis paralysis, with Boyd's rule of thumb: seventy per cent right now beats a hundred per cent perfect next week. Claude asks which two or three options are realistic, which gives the highest tempo, which keeps options open and which closes them off, what the least bad fall back is and what the smallest possible commitment is that you can still learn from. At Act everything is about closing the loop: the first step today, the measurement that proves success or failure, who needs to be involved and when the loop runs again.

There is also a dedicated section on tempo as a weapon, asking how you shorten the Orient phase without losing quality, what rhythm you keep to, such as a daily check and a weekly pivot, and how you push decision authority lower into the organisation. Boyd called that Auftragstaktik, mission command: hand over the intent, not the instruction. And it lays down a fixed output structure, the five parts you saw in FIG.01: Observe observations in three to five bullets, the Orient analysis, a maximum of three Decide options, the Act plan with a measurement point and a deadline, and a tempo check. Under the principles is also Boyd's order of things: people, ideas, hardware, in that order. Tools always come last.

How an instruction file like that is put together and how to write one yourself is explained separately from this skill in writing SKILL.md NL.

Three things shift in practice. First the language: panic states conclusions, the loop states observations and assumptions. Second the number of options: under pressure you see one, or exactly twenty, the skill forces it down to two or three realistic ones. Third the ending: an ordinary brainstorm ends with a good feeling, a loop ends with a measurement point and a date to look again. In diff form, with lines close to the example in the file:

The second example in the file shows the same shift on a smaller scale. Someone has to choose between two features for their app and does not know which one. Then a short loop follows with the accent on Orient: what assumption do you have about what clients want, what does data say against gut feeling, which option gives the fastest learning moment. And then the sentence that sums up the whole skill: with two features the question is rarely which one is better, but which one you can disprove fastest.

After that, using it is simple: describe the situation as it is, mess and all. What you know, what you suspect, what has just changed. It sorts out what is hard and what is soft itself, and asks about what is missing. A step by step guide for every environment is in installing Claude skills NL.

The file lists its own limits honestly, and there are five. The loop is a thinking framework, not a procedure with boxes to tick, Boyd never meant it as four compartments. It works poorly for questions with a long feedback loop, such as multi year investments: a strategic framework fits better there. Tempo is not always the winning variable: in heavily regulated environments, with safety critical systems and in legal processes, care matters more than speed. The popularity of the model is greater than the empirical research behind it. And deciding fast raises the chance of mistakes, which is acceptable as long as the mistakes are small and reversible.

That last limit deserves its own sentence, because there is a second skill behind it. For big and irreversible steps you deliberately slow the loop down, the file says. But how do you know if a step is irreversible? There is a separate thinking framework for that, Jeff Bezos's Type 1 and Type 2 framework, and that sits in the Two-Way Door Decider skill NL. The two complement each other exactly: the door check tells you how much care a decision deserves, the loop tells you how to stay sharp under tempo.

And the boundary the frontmatter draws itself: are you looking for the cause of a problem that has already happened, then you are in root cause territory and a 5 Whys analysis makes more sense than a loop. The OODA Loop looks forward to the next move, not back to the source of the last one.

The first phase of every loop is Observe, and that applies to your visibility too. Many businesses decide on their marketing by gut feeling, without a single hard observation about where they stand. The free SEO scan gives you that observation in a few seconds: hard, measurable and without judgement, exactly as Boyd wanted it. In the guide eight decision models compared NL you can see how this model relates to the others. And if you want to talk through what else AI can do for your business, from single skills to full automation, we simply do that in a conversation.