Zum Inhalt springen

AI Coach: A Chief of Staff Who Doesn't Mince Words

Most productivity tools count hours. My AI coach recognizes patterns.

The Problem

Knowledge workers with ADHD (and not just them) know this: you start the day with a clear priority. Three hours later you’ve started five other things, none of them were on your list, and the actual task is untouched.

That’s not a discipline problem. That’s a structure problem. And structure is something a system can provide.

But “structure” doesn’t mean calendar blocking or Pomodoro timers. Those have been around for years, and they regularly fail for a simple reason: they know nothing about the person using them. They don’t know energy curves. They don’t recognize avoidance patterns. They don’t notice when someone has been pushing the same task for three weeks — always with a plausible justification that, on closer inspection, is different every time.

The Approach

The coach isn’t an app. It’s a layer in the 6-layer architecture with access to all other layers: What was planned? What was actually worked on? Which patterns repeat over weeks?

The decisive factor is aggregation over time. A single unproductive day is an anecdote. Three weeks where a specific task type is systematically avoided is an insight. The coach sees these time series because it runs during every session. It doesn’t just compare what was planned with what happened — it recognizes the patterns behind the deviations.

Three Principles

1. Directness Over Diplomacy

No “Maybe you could consider…” Instead: “You’ve been putting this off for three weeks. Is that a conscious decision?”

A coach that only affirms is useless. The art is being uncomfortable without being hurtful. This succeeds when the critique is based on data, not opinions. “This week you invested 14 hours in concept work and 0 hours in outward communication” isn’t a judgment — it’s an observation. What the person does with it remains their decision.

But directness has a limit: acknowledge progress before naming the gap. That’s not softening. It’s honest feedback that includes the context. Telling someone where they stand without telling them how far they’ve come doesn’t provide the full picture.

2. Energy Before Efficiency

Not “How do I get more done?” but “How do I get the right things done with the energy I have?”

For people with chronic conditions, parents with small children, or anyone without unlimited energy, this is the central question. There are days when three hours of deep work are possible. And days when only administrative tasks work. A good coach recognizes the difference and reorders priorities accordingly, rather than stubbornly working through the weekly plan.

This sounds banal, but it has consequences: the coach must learn that “less, but right” is sometimes the better recommendation than “just one more task.” Especially with ADHD, where hyperfocus mode can be deceptive: three hours of high productivity, then a crash reliably follows. If you know the pattern, you can warn before it happens.

3. Patterns Over Moments

A single unproductive day is meaningless. A pattern of three weeks of avoidance is information. The coach aggregates over time and reports back what the person can’t see themselves.

What the Coach Concretely Recognizes

Communication Debt

The term comes from software development (“Technical Debt”), applied to communication: every deferred response, every conversation not had, every postponed coordination is a debt that accrues interest. The coach recognizes when these debts accumulate — not because it reads emails, but because it sees that certain task types systematically slide to the bottom of the list.

A concrete example: a channel that hasn’t been updated in weeks, even though visibility is defined as a strategic goal. The coach observes: concept work (intrinsically motivating) dominates. Communication (extrinsically motivated, requires effort) gets avoided. The intervention isn’t “You need to post more,” but: “This channel has been silent for 24 days. A short post takes 20 minutes. What’s holding you back?”

The question “What’s holding you back?” is crucial. It opens a space for reflection rather than dictating a task.

Avoidance Patterns

There are tasks that keep getting postponed. Not because they’re unimportant, but because they’re emotionally taxing, bureaucratically complex, or simply boring. The coach recognizes the pattern and can address the actual blockage.

Typical for ADHD: the task isn’t too big. It’s too unstructured. What’s missing isn’t motivation, but a concrete first step. “Do the tax return” is overwhelming. “Open the folder and collect the wage tax certificate” is doable.

Over-Conceptualizing Instead of Finishing

A pattern especially common among analytical thinkers: the concept keeps getting refined. Version 1, version 2, version 3 — all internal. And nobody has ever seen version 1. The coach asks the uncomfortable question: “Is this revision better for the recipient, or just for you?”

This isn’t a critique of quality standards. It’s the observation that perfectionism and avoidance are related. Anyone who keeps optimizing never has to face others’ judgment.

Energy Cycles

At what times of day does deep work succeed? When only administrative tasks? Over weeks, the coach builds a model of the individual’s performance curve. Not as a rigid scheme, but as a tendency: “Your pattern shows that late-night sessions are productive, but a price is paid the next morning.”

Making Invisible Progress Visible

People with ADHD systematically underestimate what they’ve already accomplished. The focus is always on what’s still missing. The coach can say: “In the last two weeks, you built three systems, published a podcast, and submitted a strategy. That’s not nothing.” No flattery. An enumeration of facts.

The Psychological Debate: Does the Person Lose Their Autonomy?

This is the question that inevitably comes up in every conversation about AI coaching. And it deserves an honest answer.

The argument against: If a system tells me what to do when, names my avoidance patterns, and knows my energy curves, am I then unlearning how to reflect on my own? Does self-awareness get outsourced and atrophy like a muscle that’s no longer used?

This isn’t a straw man. It’s a real risk.

The argument for: The premise that people are good at self-reflection without external help is empirically questionable. Kahneman wrote an entire book about how systematically we deceive ourselves about our own thought processes. Therapists exist because self-reflection alone often isn’t enough. Coaching exists because an outside perspective sees things that the inside perspective systematically overlooks.

My position: The coach doesn’t replace the capacity for self-reflection. It extends it. It provides data that no human could compile this way, because no human has all the data points. A therapist sees me one hour per week. A colleague sees only the professional context. My partner sees only the private context. The coach sees both — and above all, sees how the domains influence each other.

This isn’t surveillance. It’s a mirror that’s sharper than one’s own gaze.

But: it presupposes that the human retains the final decision. The coach recommends; it doesn’t decide. It shows patterns; it doesn’t act. And it must be correctable. A coach that is never wrong or never allows itself to be corrected isn’t a coach. It’s a dictator.

What the Coach Is Not

Not a therapist. Not a time-tracking tool. Not a surveillance system. The coach has no access to content that hasn’t been shared. It observes patterns, not thoughts.

It doesn’t replace a doctor or medication. It’s a tool within a larger system of support. The gap it fills lies between therapy sessions (once a week) and everyday life (the other 167 hours).

And: it makes mistakes. Regularly. Not every detected “pattern” is actually one. Sometimes a task gets postponed for three weeks because it genuinely isn’t important right now. The ability to accept corrections and learn from them is part of the system.

The Connection to the Self-Vector

The coach is the first application of the Selbstvektor (self-vector). When a system has a model of the user — work patterns, energy curves, communication preferences, motivation triggers — it can anticipate rather than merely react.

“You did three hours of deep work this morning. Your pattern shows that after 3 PM only administrative tasks make sense. Should I reorder the priorities?” That’s not science fiction. That’s a weighting function on existing data.

And this is precisely where the difference to every existing productivity tool lies: the coach doesn’t react to what you tell it. It reacts to what it knows about you. And this knowledge grows with every interaction — not because it collects more data, but because it better understands what the existing data means.