Why do managers do this? Are they afraid that somebody will make them look bad? Do they get a power kick out of being the boss? Do they call themselves detail-oriented and insist that they know what is best?

Whatever the reason, the result is the same. Responsibility travels outward. Authority stays in the centre. People are told that they own an outcome, but every meaningful decision still has to return to the manager.

Responsibility without authority is less than useless. It does not simply fail to help. It leaves somebody accountable for a result they were not allowed to shape, while the manager keeps every decision and becomes the bottleneck.

There is another approach. If you give somebody responsibility, give them the authority needed to act within a clear framework.

That does not mean abandoning supervision. The manager defines the framework and its boundaries, remains available to guide and challenge, and serves as the backstop when one is genuinely needed. Inside that framework, the person responsible is free to act and make decisions.

Then watch what happens. People stop waiting for permission and start producing outcomes. They take ownership of the product, process, department, or whatever else they have been asked to run. Quite often, they surprise you.

There will be mistakes. Sometimes, especially at the beginning, you could have done the work faster or better yourself. That is the price of building capability. If you reclaim authority after the first imperfect decision, you teach people to wait for instructions rather than own the result.

With time, people become more confident, enjoy their work more, and take real ownership. Their success is also yours. Meanwhile, you are no longer carrying every detail or blocking every decision. Your stress level may fall. You may even become a nicer person to be around. That is a positive feedback loop that actually deserves the name.

Two cautions.

Some people do need close supervision. In my experience, however, they are not the majority. Take that into consideration, but do not use it as an excuse to design every role around distrust.

And if delegating authority sounds insane or dangerous, ask the more basic question: why were you willing to delegate responsibility in the first place?

The app is for the parts of Zettelkasten work that happen away from the desk. While reading a book at home or in a café, you can capture an idea as an Inbox note or begin a source-bound literature note before the thought disappears.

Connect the app to your existing Other Memory Zettelkasten in Files or iCloud Drive and those captures become part of the same file-first vault you use on the Mac. If you create a new source, Other Memory prepares a pending bibliography record for later reconciliation on the Mac and handoff to the bibliography manager you already use.

You can also search your placed Zettels by address, title, or text, read them on your iPhone, and follow the links you have created.

Other Memory for iPhone is free. You can download it from the App Store.

At first I thought it had to be a joke dreamed up by someone trying to generate views. It turns out it is quite real. For a moment I wondered whether the AI companies had invented it themselves. It would certainly be good for token sales. But no conspiracy is required. A bad metric is enough.

Tokens measure consumption. They do not tell you whether anything useful happened. Counting them as productivity is activity dressed up as outcome. It is throwing things at the wall and rewarding whoever throws the most.

That led me back to another management favourite: thinking outside the box. We repeat the phrase so often that it has become a mantra rather than an instruction. Tokenmaxxing is a useful reminder that something does not become intelligent merely because it is novel or excessive.

But what is the box?

If we define it as our area of competence—our knowledge of the work, its constraints, its failure modes, and the people affected by it—then thinking outside the box in a positive sense means making that box bigger. It means connecting genuine expertise to new ideas, tools, and possibilities.

That only works if you understand what is already inside.

What if you do not yet understand your own box? What if you do not know how the work really happens, where it breaks, or why experienced people make the decisions they do? What exactly are you thinking outside of? In that situation, “outside-the-box thinking” can become a substitute for learning the job.

I have met plenty of supervisors and operational managers who know their part of a business in extraordinary detail. They know where the work jams, where the official process diverges from reality, which shortcuts are dangerous, and which apparently small problem is costing everyone hours. These are prime candidates for expanding the box.

Those are the people I want to tell: I will invest in you. I will help you learn how to use AI inside your area of expertise—not to replace what you know, but to extend what you can do with it. Once they get the hang of it, step aside and watch them run.

Unfortunately, I have also met CXOs whose appetite for thinking outside the box is far greater than their appetite for understanding what is inside it. AI gives them the ideal shortcut. It can be mandated, announced, purchased, and implemented by fiat. Everybody is told to transform, and outcomes are expected to appear.

Three months later, somebody is asking how the organisation managed to spend, say, $250,000 on API tokens and get bupkis in return.

That is not thinking outside the box. It is trying to skip the box.

AI is most useful when it gives leverage to real competence. Before asking an organisation to think outside the box, find the people who genuinely understand what is happening inside it. Invest in them first. Help them enlarge the box.

Then watch them run with it.

Just a thought.

If you have chatted with ChatGPT, then technically you have used AI. If it has become your helpful new friend, fine. But that does not, by itself, mean you know how to use AI for work. The chat interface is the front door. It is not the whole house.

You begin acquiring practical experience when you use AI to do something productive for which you remain responsible:

• as a research assistant for your PhD;
• as a development partner while writing software;
• as an analysis system for production data or weaknesses in your operations;
• as an artistic partner for creating visuals or motion pictures;
• as a way of managing work that you would otherwise fail to manage; or
• as part of your marketing analysis and the delivery of actual marketing projects.

That is when the friendly conversation becomes a working relationship—and the awkward questions begin. Which model actually suits this job? How much context does it need? How many tokens will the project consume, and what will that cost? Should you use OpenAI or xAI, one of the other Western providers, or perhaps a Chinese model that fits the task better? Can the tool plan the work, preserve that plan, and show you when it has wandered away from it?

If you have never used the planning functions in your chosen tool, compared models, watched a token budget, or abandoned one model because another fitted the job better, then you have barely started.

The most important lesson is that an AI answer is not the final answer. AI gives you options, drafts, analyses, and possible routes. It can also confidently invent things, miss the point, and sprint down a rabbit hole unless you control it. You have to set the scope, demand evidence, establish checkpoints, build guardrails, and know when to stop it.

So please do not put “AI” on your CV simply because ChatGPT has become your new helpful friend. Put it there when you can explain what you used AI to accomplish, which tools and models you chose, how you verified the work, where the system failed, and how you controlled its cost, scope, and risk.

That is experience. Everything before it is acquaintance. Acquaintance can be useful, but it is not the same thing.

Most of my work is on that moment: what Ipoku should say, what it should leave unsaid, and when the phone has to stop being part of the conversation. The interesting thing is not the machinery. It is whether the person looks up.

I am also making sure that both apps work with ordinary Markdown files, and that the Mac can export the Zettelkasten into a physical form that continues on paper. That boundary matters to me. Other Memory should help me use the method without making the method dependent on Other Memory.

I am checking that the Card, the rounds, the board, the decision trail, and the thirty-day trial describe one method from beginning to end. The diagrams have to explain something. The notes have to earn their place. Most importantly, the finished book still has to sound like the person who wrote it. If a paragraph begins to sound like a consultant presenting a maturity model, it goes back on the bench.

The work now is making each card useful on its own. If a manager has to study the application before using the card in a meeting, the card has failed. RDAC should supply enough doctrine to move the decision and then get out of the operation's way.

I am working through the definitions, examples, Cards, rounds, boards, diagrams, and the thirty-day trial. The purpose is not to make the book gentler or turn it into another management framework. It is to make the method clearer where clarity helps and keep the direct voice where directness matters. Better English is welcome. Consultant English is not.

A lot of writing about digital Zettelkasten treats Folgezettel as a historical accident, or says the addresses stopped mattering once software could link anything to anything. I disagree. The address records where a thought entered a line of development. Remove that and you may still have a useful notes application, but you have removed an important part of the method I want to preserve.

One permanent card is therefore one ordinary Markdown file with a meaningful address. Images, references, and indexes may help, but they don't get to become a second hidden truth. If the application disappears, the thinking must still be there.

The finished first release has Catholic, Protestant, and Orthodox reading structures, daily progress, streaks, and a guided way to build a more personal reading pattern. It also has fourteen interface languages, which is a useful reminder that a short English button can become a surprisingly large international problem. The app still has one modest job: tell you where to read next, remember where you stopped, and leave the actual reading to you.

The idea ran ahead of the practical technology. It no longer does. We finally have the technology to do something about it, so I have returned to Ipoku. The point has not changed: use the phone to make the first real-world interaction less blind, without turning people into public profiles or exposing the private information behind the signal.

I have started RDAC to deal with that problem. It borrows the useful part of military doctrine—not the theatre, uniforms, or desire to call the office a battlefield, but the discipline of recognizing the situation, deciding with incomplete information, acting, and checking what reality did to the decision.

The first version begins with the Catholic canon, daily progress, and a simple question: what should I read next? The app should answer that and then get out of the way. It is meant to sit beside a Bible, not become a smaller and more distracting substitute for one.