MetaMemoryWorks lets you build complex assistant systems from small, manageable building blocks.
Each block is a module of its own, supporting you in a specific part of everyday life: nutrition, sleep, training.
Or in more focused situations: at work, while writing, applying for jobs, or learning something new.
Each module focuses on one concrete problem.
By working with clear files, explicit rules, and memory that persists, you stop having to start from scratch over and over again.
You can use modules on their own.
Or you can combine them gradually, once you notice that it actually makes sense for you.
And for private, non-commercial use:
if you enjoy a bit of tinkering, you can even build your own modules 🙂
There’s no guitarOS, dogOS, gardenOS, or philosophyOS yet —
but once you understand how the systems work,
creating something like that often takes just a few minutes.
🧩 scanOS — Ingest & Structure
scanOS makes visual information usable.
Anything you can see can be captured, described, and processed with scanOS —
for humans, or for other modules.
scanOS can read, for example:
- screenshots from apps (e.g. fitness, health, or tracking apps)
- photos
- visual documents and exports
The content is structured and output
either as a human-readable description
or as a machine-readable JSON description.
This makes it easy to store screenshots from fitness apps in a clean, consistent way
and reuse them later with modules like trainingOS.
scanOS works especially well as the start of a pipeline:
it ensures that visual information actually arrives,
instead of sitting around as unused images.
→ View scanOS
🗂️ noteOS — Memory for Thoughts
noteOS makes thoughts durable.
Everything you write or dictate can be captured, organized,
and revisited over time with noteOS.
For example:
- thoughts
- notes
- ideas
- drafts
- short or longer logs
Entries are not stored as loose text fragments,
but logged persistently and in structured form.
Nothing simply disappears. Things can be found again later,
combined, or analyzed.
For many people, noteOS quickly becomes the backbone of their system:
the place where thoughts don’t get lost,
but are allowed to grow over time.
→ View noteOS
🏋️ trainingOS — Training with Memory
trainingOS is a training assistant that actively thinks along.
It helps you build training in a way that fits you —
your everyday life, your motivation,
the equipment you have,
and what’s realistically possible right now.
trainingOS doesn’t plan training in the abstract.
It starts from what’s actually there.
It guides you through sessions, remembers what happened,
and uses that knowledge to make suggestions
that feel reasonable — not random.
trainingOS uses AI to actively optimize your training.
The difference:
the AI doesn’t operate in a vacuum,
but on the basis of memory — your training history.
trainingOS remembers how you train,
what you set out to do,
what worked — and what didn’t.
Based on this, trainingOS can be individually adjusted:
plans evolve, goals are refined,
load and volume can be adapted on the fly.
The result is not a rigid training plan,
but a system that responds to real conditions —
to progress just as much as to breaks, fatigue, or injury.
trainingOS is not a fitness app and not a point system.
It doesn’t judge you — it supports decisions.
Not: “This no longer fits the plan.”
But: “Then we adjust the plan.”
→ View trainingOS
🍽️ nutritionOS — Nutrition with Memory
nutritionOS helps you cook with what you actually have.
It remembers what’s in your pantry,
which recipes you use,
and what works well for you.
Optionally, it can also keep track of supplements.
Everything else builds on that.
nutritionOS suggests recipes, adapts them,
and accompanies you while cooking —
not based on idealized plans,
but on what’s really available.
It can estimate energy needs,
put calorie expenditure into context,
and suggest meals that fit those needs.
In combination with trainingOS,
this can also reflect the day’s physical activity.
This makes it possible to steer nutrition
without constant counting or feeling hungry.
What works best develops step by step —
nutritionOS learns along with you.
Along the way, it can help you:
- plan enough fruit and vegetables throughout the day
- balance macronutrients
- keep an eye on micronutrients
nutritionOS doesn’t judge and doesn’t moralize.
It doesn’t label food as “right” or “wrong”.
You can use it to cook very healthily.
You can use it to cook extremely well.
And it can support you when losing weight — or gaining it.
nutritionOS is not a diet program.
It’s an assistant system
that makes nutrition practical, flexible,
and surprisingly relaxed over time.
→ View nutritionOS
🔜 Coming next
sleepOS
đź§ Architecture & Method (for the curious)
All MetaMemoryWorks modules are built on a shared memory and state architecture.
The basic idea is simple:
AI works better when it doesn’t have to reinvent itself every time,
but can build on a clear, reconstructable state.
To achieve this, MetaMemoryWorks uses, among other things:
- persistent memory in plain, ordinary files
- explicitly reconstructable state instead of implicit chat history
- logs and documents as explicit memory building blocks
- modules that remain independent while still working together
- an approach that isn’t tied to any specific model or system
This architecture forms the common foundation of all modules —
and is open enough to support your own ideas, experiments,
or entirely new assistant systems.
It’s particularly interesting for:
- people who enjoy building and tinkering with systems
- teams and organizations
- research and education
- technical decision-makers and system architects
If you’d like to dive deeper, you can find the technical documentation here:
→ View the MetaMemoryArchitecture repository
đź’š Support
MetaMemoryWorks is free for private use.
If the modules help you in everyday life,
if they save you time, create clarity,
or are simply enjoyable to use,
supporting continued development is greatly appreciated.
It helps build new modules,
maintain existing ones,
and keep MetaMemoryWorks open and accessible in the long run.