Trithemius

How to live when your life’s work turns into a toggle in Settings

Hey.

Let me tell you a story.

It is winter in the late 1400s. Snow is coming in sideways over the Rhine. A young abbot walks into a cold stone room lit by candles. Rows of monks are bent over desks, copying words that have already been copied for a thousand years.

His name is Johannes Trithemius.

Books are still holy objects. A full Bible is worth as much as a small house. The work of your hands is visible on every page. Your handwriting is your fingerprint.

Outside the monastery walls, something new has arrived in Europe.

A machine that can do in a single day what a room of monks could not do in a year.

The printing press.

Trithemius does something that feels very human and very familiar. He writes a book called De laude scriptorum - In Praise of Scribes - defending the value of hand-copying manuscripts, even as he quietly uses printing to distribute his arguments further.

Midjourney: “Trithemius working hard”

He knows the press is powerful. He calls it a marvelous art. He still argues that the scribe who copies by hand is closer to God, closer to the text, closer to meaning.

Sound familiar?

You are Trithemius. Your job is to hand-copy manuscripts. The AI is the printing press.

Except this time, the press can talk back.

---

Intelligent Economics

There is a new frame that I cannot get out of my head.

It is called Intelligent Economics.

The core idea is simple and uncomfortable:

The rise of scalable, non biological intelligence flips our economic logic, turning most human cognitive labor from positive to negative value in a very short window.

The author calls this the Intelligence Inversion - the moment when silicon becomes better, faster, and cheaper at most forms of thinking work than carbon. The claim is that we are not only in a productivity boom. We are at the end of scarcity based economics altogether.

For most of history, value came from moving atoms. Land. Grain. Steel. Oil.

Then we added bits. Financial derivatives. Software. Advertising. SaaS valuations.

Now the scarce thing is not information. It is sorted information. Correct information. Useful information. Reliable information.

The frame that stuck with me is this:

Value is the act of sorting.

Sorting is what reduces entropy. Turning chaos into order. Noise into signal. Raw data into a decision you can actually use.

• An analyst who distils fifty reports into one clear slide is sorting.

• A founder who picks one crazy idea out of fifty mediocre ones is sorting.

• A recruiter who filters a thousand CVs into three offers is sorting.

In this lens, intelligence is not a number on a test. It is the capacity to sort reality into useful patterns.

Artificial intelligence is a sorting machine that can scale.

That is where the Thousand Day Window comes in.

Once general-purpose AI becomes widely available, most human cognitive labour becomes economically negative within one to three years, because a slower, more expensive sorter attached to a fragile human ego cannot compete with a faster, almost free sorter that runs twenty-four-seven.

You can feel this already if you are honest.

You send a document to a colleague to summarise.
They send it back in two days.

You paste the same document into an AI model.
You get three versions of the summary in twenty seconds, plus ten alternative angles and a slide outline.

Who generated more value, per unit of cost and time?

That gap between human effort and machine sorting is going to widen, not shrink.

---

MIND scoreboard

Intelligent Economics proposes a new dashboard for value creation called MIND:

• Material Capital

• Intelligence Capital

• Network Capital

• Diversity Capital

Think of it as a new scoreboard for the post-scarcity economy.

Material is obvious. Land, factories, chips, energy.

Intelligence Capital is less obvious. It is not degrees or certificates. It is access to and mastery of intelligent systems - models, agents, toolchains - plus your own ability to dance with them.

In

I wrote that the real frontier is not being an expert in one frozen domain. It is being able to generate temporary expertise on demand with AI as your exocortex. That is just another way of saying you are compounding your Intelligence Capital.

Network Capital you already understand if you have ever raised money or closed an enterprise deal. Who trusts you. Who answers your messages. Whose Slack you can DM at 23:00 and get a serious answer.

Diversity Capital is not a corporate training session. It is the range of perspectives, modalities and experiments you actually run. Different data, different cultures, different model families, different environments. In a world of massive optimisation, monocultures are how you die.

In

I argued that corporate structures will be forced to unbundle into smaller, more effective units because large organisations cannot adapt fast enough once AI blows up their task graph. Intelligent Economics gives you the mechanics underneath that story.

If your company is all Material and no Intelligence or Diversity, you are a factory in 1910 right before electrification hits.

---

Three futures: Feudalism, Fragmentation, or Symbiosis

From this starting point, Intelligent Economics sketches three possible futures.

Let us translate them into normal language.

1. Digital Feudalism

One path is that a tiny number of actors own the sorting machines.

• They own the models.

• They own the data.

• They own the chips and the power contracts.

• They own the distribution.

Everyone else rents intelligence like medieval peasants rented land.

You already feel this when you realise your entire business runs on one provider’s API, and they silently change the rate limits and pricing; your margins disappear overnight, while your roadmap turns into their changelog.

Your life becomes a stack of subscriptions inside a few super apps. You get comfort, convenience, and a perfectly optimised feed in exchange for your agency. Human purpose is outsourced to engagement metrics.

When you catch yourself thinking,
“I will just wait for my company to give me a license,”
That is a feudal mindset. You are a tenant, not an owner.

2. The Great Fragmentation

Another path is a patchwork of incompatible systems.

Every nation, every city, every corporation uses its own models, standards, and constraints.

Open weights here, closed systems there, blacklists and whitelists everywhere.

Innovation happens, but in silos.

Routing intelligence across borders becomes as messy as trying to mail a parcel in Europe before there was an integrated system. Everyone has their own rules, forms, and adapters.

If Sub Agents was about the explosion of specialised AI workers, The Great Fragmentation is what happens when they cannot talk to each other.

3. Human Symbiosis

The third path is the only one where we do not end up as overqualified content moderators for machines.

Call it Human Symbiosis.

The idea is that we consciously choose an Alignment Economy where we pay not just for output, but for alignment with human purpose.

That is where the Dual Currency System comes in.

You still have a normal economic currency for matter. Energy. Chips. Food. Housing.

Additionally, you track and reward a second currency for aligned intelligence work.

• Not “time spent in front of a screen”.

• Not “number of prompts written”.

Things like:

• How many people did this product actually help

• How robust is this system to misuse

• How resilient is this community when the model fails

Why does this second currency become inevitable?

Because without it, human purpose becomes an externality. And in optimisation systems, externalities get crushed. Anything you do not measure becomes invisible to the objective function. Anything invisible gets traded away for a small bump in click through rate or margin.

The Dual Currency System is not a utopian add on. It is a safety belt for civilisation. Either we price human purpose in, or the sorting machines will price it out.

You can call this second currency whatever you want. Meaning Capital. Alignment Credit. Human Dividend.

The important part is that you do not leave it implicit. You measure it. You reward it. You put it on the scoreboard next to revenue.

---

AI: Trithemius gets a talking printing press

All of this would be abstract if the press were still in the lab.

It is not.

The new printing press is not a single product release. It is the whole category.

GPT 5.1, Claude, Gemini, local models on your own GPU - these are just different brands of the same civilisational machine.

They do three things at scale:

• Compress
  Take too much information and squeeze it into something you can actually hold in your head. Summaries, briefs, bullet points, decks.

• Expand
  Take a tiny seed - a sentence, a sketch, a feeling - and grow it into drafts, code, designs, scenarios.

• Coordinate
  Take many moving parts - tools, APIs, calendars, documents, people - and route information between them.

That is what the printing press did for text.

It took ideas that were stuck in one place, one brain, one script room, and compressed, expanded, and coordinated them across continents.

The difference now is speed and intimacy.

The press sits on your phone. It talks in your language. It mimics your tone. It remembers yesterday’s conversation.

The identity of the model matters less than your relationship with it.

Right now most people relate to AI like this:

• A search engine with better vibes

• A homework machine they half-trust

• A toy that occasionally spits out something impressive

If you are Trithemius, that is the same as using the printing press only to stamp your grocery list.

The upgrade is to treat AI as a universal press for your own manuscripts:

• your knowledge

• your processes

• your reputation

• your relationships

• your taste

Whether you are using GPT 5.1, another model, or a local setup, the question is:

What am I actually printing with this power, day after day?

Once that clicks, the specific model is just a choice of printer brand.

Midjourney: Trethemius as a Boy

---

GPT 5.1 is one example of the new press

Within this broader context, GPT 5.1 is just one example I personally use frequently.

It is interesting to me not because of one benchmark jump, but because of the direction:

• More conversational, less robotic

• Modes for quick answers or deeper thinking

• Personality controls so you can tune tone and style without restarting every time

This matters because it reduces the friction of using AI as a daily collaborator, rather than a tool you open only once a week to experiment with.

In AI Generalist I wrote that the real competitive edge is not access to models. It is AI intuition - your ability to dance with the system, rather than barking commands at it.

GPT 5.1 is a better dance partner.

It remembers the rhythm of the conversation. It can switch between roles midstream - research assistant, ruthless editor, therapist, product manager - without losing the thread.

The newest prompting guides read less like manuals and more like choreography boards.

I remain, and will continue to be, very bullish on AI.

Not in the abstract “this will change everything” way, but in the practical “I live inside these systems every day, and they keep expanding what I can build” way.

Right at the moment I use AI heavily to:

• Turn half-formed fragments into testable ideas

• Stress test acquisition theses for Belkins and Equinox

• Design and refine processes at Folderly without burning my team

• Draft and redraft these newsletters until they feel like me, not like a manual

If you read Time for Idea People and Ideation, you already know my position.

The cage is open. The question is not whether AI will replace you. The question is whether you are willing to become the kind of human that AI cannot easily compress.

---

Choose your path with the press.

Different people read this newsletter for different reasons, so let us make this concrete.

If you are an employee or knowledge worker

Your job is mostly sorting already.

Emails, meetings, reports, tickets, clients.

Use AI to:

• Turn every recurring task into a mini playbook
  “When an X email comes, I usually do A, B, C.”
  Please write it down. Feed it to your model. Let it generate first drafts.

• Protect your deep work.
  Use AI to draft agendas, summaries, and follow-ups so that meetings become shorter and rarer.

• Show your manager before they ask
  Once a week, use AI to compress your work into one clear update:
  What changed, what blocked you, where AI helped. Become visibly higher leverage.

You are not trying to be “the AI person”. You are quietly becoming the person whose output suddenly doubled without burning out.

If you are a founder or manager

Your leverage is where AI hits the profit and loss.

Use AI to:

• Redesign one process per month
  Take a core process, such as hiring, onboarding, outbound, support, or reporting, and run it end-to-end with AI in the loop. Measure hours saved and error rate.

• Stress test decisions
  Before big calls - hiring, pricing, funding, firing - ask your model to steelman the opposite choice and show you what future you would criticise.

• Build an “AI first” culture, not an AI department
  Encourage everyone to document their prompts and wins. Make it normal to ask, “How would we do this if a smart model sat in the middle?”.

You are not buying a tool. You are changing the unit economics of thinking inside your company.

If you are a creator or freelancer

Your raw material is attention and trust.

Use AI to:

• Record everything, forget nothing
  Turn calls, notes, brain dumps into searchable archives with summaries and tags.

• Increase frequency without lowering quality
  Let AI draft first passes and research, then you spend your energy on voice, taste and honesty.

• Productise your brain
  Use AI to help turn your best ideas into ebooks, courses, templates, checklists that can earn while you sleep.

You are not competing with AI content. You are using AI to remove all the friction between your brain and the people who want to hear from you.

If you are a student or early career

You are not late. You are right on time.

Use AI to:

• Turn it into a tutor
  Ask it to explain things at three levels: child, student, and practitioner. Make it quiz you. Make it show you where you are actually weak.

• Simulate futures
  “Show me a week in the life of a good product manager, nurse, founder, or engineer in 2030.”
  Try out careers before committing years.

• Build a visible portfolio
  Use AI to help ship small projects you can show - scripts, analyses, essays, tools - instead of just collecting certificates.

You are not trying to win the old game. You are training for the game that will exist in 5 years.

You are Trithemius.

Back to our monk.

Imagine Trithemius sitting in his scriptorium, watching the first printed books arrive.

Clean lines. Identical pages. No coffee stains, no marginal doodles, no tired mistakes.

He loves books. He fears losing the soul of the work.

So he writes in praise of scribes.

We are doing the same thing today.

Every time a lawyer tells you, “No serious client will accept a contract that was drafted with AI.”
Every time a professor says, “Real students write their own essays.”
Every time a copywriter posts, “My clients come to me for my voice, not for robots.”

It is Trithemius energy.

I get it. I feel it.

But here is the twist.

Trithemius also used the press. He had his own books printed. He expanded his abbey’s library from forty volumes to two thousand, combining manuscripts and printed works.

He was both defender of the old craft and architect of the new infrastructure.

That is the job now.

You are not supposed to choose between hand copying and printing.

You are supposed to:

1. Stop defending the quill
   No one gets a medal for doing slow work that a machine can do better, only because it feels artisanal. Keep the handwork where it genuinely adds meaning, not where it adds delay.

2. Teach the press what matters
   Most “AI replacement” stories are really stories of humans who refused to encode their judgment into prompts, workflows, playbooks, and training data. Your taste and ethics, and domain knowledge have to be legible to the machine, or they do not exist.

3. Write scriptures for machines
   Trithemius’ monks copied scripture so that humans could read it. You are now writing the scriptures that machines will read - policy files, evaluation suites, reward models, and datasets. Even your Slack culture and Notion documents are influencing how your organisation will utilise AI.

If Intelligent Economics is roughly correct, then in this Thousand Day Window your options look like this:

• Stay a scribe and get slowly automated

• Become a printer operator for someone else’s machine

• Or become the person who decides what the machines are printing and why

The last category is small and very under-subscribed.

---

How to use this, personally

Regardless of which group you fall into, you can think of the next thousand days as your personal Trithemius window - the period where you decide whether you are the scribe, the printer operator, or the person who decides what gets printed.

Use MIND as a personal checklist, not a philosophy:

• Material - Each quarter, ask: what part of my income still depends on pure time for money, and how do I convert at least 10 percent of that into productised or asset-backed value.

• Intelligence - Treat your main AI model as your second brain: every substantial decision, document, or deal passes through it at least once for stress testing or expansion.

• Network - Each month, add one person to your circle who is actually building with AI, not just talking about it on stages. Curate for operators, not influencers.

• Diversity - Deliberately feed your models and your mind new data sources: other languages, other industries, weird datasets, non-obvious problems, and keep a small log of what surprised you.

If you are reading this newsletter, you are already ahead of the average Trithemius in your industry.

The next step is not “wait and see”. The next step is to act as if the Thousand Day Window is real, even if the exact number is wrong.

As Main Character I wrote:

The universe keeps dropping hints. Your only real job is to stop pretending you cannot see them.

Right now, the hint is loud.

Models are getting smarter, warmer, and cheaper. Compute is getting concentrated. Regulation is trying to catch up. Your calendar is probably already full of tasks that a well-set-up AI agent could do better.

You can be the monk complaining about paper, or the abbot who uses the press to fill his library.

Your choice.

The Trithemius Playbook

This is the operator layer if you actually want to use all of this in your work.

1. A personal MIND audit you can run tonight

Block 45 minutes. Open your main AI model in a fresh chat, in a mode that favours deeper thinking over instant replies. Paste this prompt and work through it honestly:

You are an elite operator coach. Help me run a brutally honest MIND audit on myself.
Ask me focused questions in four blocks: Material, Intelligence, Network, Diversity.
In each block: