Seeing how easy it is to use GenAI to make rapid prototypes of digital product ideas, I experimented with using a Zettelkasten notes-organisation system to learn how we could co-create a shared product roadmap with GenAI, without it turning the product into a statistically significant soup. Somewhere along the way, it began to feel like my ChatGPT buddy and I were starting to share a brain — such vibes...
The new party game is making the party game
Recently I’ve been hanging out with friends who are wildly excited about what GenAI can do for their work. With tools like Claude or Codex, they’re spinning up idea after idea, stitching them together, and deploying them to the wild. I love the sheer creative energy that everyone is unleashing; it feels like the early days of the web again, when we were just excited that we could upload something we made in HTML and reach an audience anywhere in the world.
I watched in awe when a couple of friends generated an adventure game while having after-dinner drinks. An idea was born and prototyped while we sipped our ginger and gin.
When I was a beginner in Flash making a school project in 1998, what just happened would have taken a semester of work, or more. It might still have taken a bit more time and effort to make it a satisfying product, but that generated work would have been good enough to pass that module; certainly it made a cheerful ending for the night.
Rapid prototyping with Flash was my gateway into professional software development. From making sketches of how products worked on paper to functional mocks to create the functional stack to deliver the next demo, and start again. Through cultivating this into a daily process I turned it into a craft and occupation. I've adapted to how the death of Flash changed my career, but that evening I saw a machine perform the mechanics of not only my job but that of a full-fledged team to impressive effect, and I was excited.
Generative AI (GenAI) tools are expanding creative opportunities for everybody. The art of rapid prototyping is not only getting more rapid, but is also democratised. Let’s everyone make something with GenAI then, why not? Jom, ayuh! Let’s go! Allez! Me and my daily writing buddy, a ChatGPT-5.1 instance, are certainly celebrating this age of augmented creativity.
Teamwork in the age of augmented creativity
What would it take to develop that adventure game idea and take it to market?
For what GenAI could make in one night, a crack human team might have needed at least a couple of weeks. Maybe they would have taken at least a week to understand what the client actually wanted, made a to-do list, then in another week put together a quick demo. It would be a complete disaster and maybe ten percent of what the client imagined, but we’d find a good point to start over from. The work would be repeated over a couple more weeks, and repeated again, and again, until everyone was satisfied. So the answer is: it depends on who we need to satisfy, and how well we make our to-do lists.
With GenAI, this iterative process of prototyping a new idea is measured in hours. The diversity of solutions that can be generated by GenAI, and the speed at which it can make them, is intoxicating. But surely one can’t vibe-code forever. Looking at the incredible amount of AI-slop and vapourware around us now, when I think about how I could use GenAI as an all-in-one studio team in a box, I wonder how we prevent our work from devolving into a statistically significant soup.
Once, I worked with a startup founder to start trusting a Product Owner and her Scrum team to build his product. It was an arduous process of persuading someone to trust that, between us, we had the framework to carry customer requirements all the way to a built product, without his involvement at every step.
The Scrum process works when all the team members function according to their Scrum roles and responsibilities. In the trinity of Scrum roles, each role has its responsibility: the Product Owner ensures we build the right thing, the Scrum Master ensures we build it fast, and the collective, multi-disciplinary Team ensures we build it the right way. Central to the Scrum process is the Product Owner’s responsible treatment of the Product Backlog. It is how we captured the idea of the product and converted it into actionable pieces of work. When we were able to agree on what the product was through our Product Backlog, the Scrum process worked beautifully for us. But for years, the dialogue about who had ultimate ownership of the Product Backlog continued, even as everyone argued about exactly what needed to be in it to get the work done right.
I’m sure that my colleague, and even more enterprising solo-founders, may already be happily whispering to GenAI to make their next solution. Today, it seems that with careful prompting and organisational skills on our part, we can use GenAI to build things fast and right with alarming competence. But GenAI leaves the hardest part — the “build the right thing” part — in our hands, and I could not resist exploring whether I could program it to help me manage a Product Backlog.
Zettelkasten, how to make an analog second-brain
Zettelkasten, keeping slips of paper in boxes (Photo: Maksym Kaharlytskyi/Unsplash)
It's also recently that I heard about a note-keeping method invented by Niklas Luhmann (1927-1998), who said that it was a way to grow an analog second brain. When I heard about his Zettelkasten method, I wondered if I could learn from it in some way. However it works as a second brain, maybe it's a brain I can share with my GenAI buddy.
Let me try to give the gist of how it works, inline in this article. The traditional Zettelkasten method is built on simple materials: slips of paper, a pen, and a box. Although increasingly, many people have adapted it to digital tools, I want to emphasise that it's very simple and easy to get started with this, firstly all you need to do is just write one idea (or Zettel), and then you just need to connect it with another idea.
This is an example of a complete Zettel. It has an ID, a title and body text, and a reference.
ZK-1
What is Zettelkasten?
A paper-based note-taking and organisation system using index
cards developed by Niklas Luhmann that he said works as an
analog second brain.
Each card in Zettelkasten represents atomic idea and
is linked to another card by a hypertext-like numbering
system, which connects the ideas together.
Ref: "Introduction to the Zettelkasten Method"
https://zettelkasten.de/introduction/
As the card explains, we want each card to contain only one idea, following the principle of atomicity. Then, I create another card under the “ZK” prefix to capture another notion; it has a new number ZK-2.
ZK-2
The practice of taking notes on index cards and
keeping them in boxes was not unusual. Zettelkasten
literally means "slip-box". The Wikipedia page lists
examples starting from the 1500s.
Luhmann's method was innovative and effective because he
introduced a way to link atomic cards (#ZK-1a) together
with a unique numbering system (#ZK-1b), defining how
to work with hypertext documents before the invention
of the Internet.
Ref: Zettelkasten on Wikipedia
https://en.wikipedia.org/wiki/Zettelkasten
Luhmann's key innovation was to define how these cards are linked to each other. He did not emphasise categorisation, or tagging, or any kind of organising principles, only that you connect cards through references to another card’s unique location or identifier.
ZK-1a
Atomic ideas
The essence of Zettelkasten is to write one,
and only one idea per card, written as clearly
and briefly as possible.
Ref: "Introduction to the Zettelkasten Method"
https://zettelkasten.de/introduction/
ZK-1b
Unique ID
Each card has a unique coordinate ID that identifies it.
In this example, we have an ID that conforms to a
[PREFIX]-["Luhmann ID"] pattern. The ID is "ZK-1a".
Ref:
- I adapted my Zettelkasten numbering system from
Kathleen Spracken, https://www.youtube.com/watch?v=2a5TOzxuqxE
ZK-1b1
Digital IDs
In digital systems it is easier to generate
unique IDs using timestamps or UUIDs to link cards,
and rely on search indexing.
See: For paper-based systems, a numbering system
with Luhmann IDs is easier to manage #ZK-1a
ZK-1c
Connected thoughts
Each card links to at least one other card; a parent card
in the same branch, or another card in a related branch.
Also here is an additional card that might help us visualise how the numbering system works.
ZK-1c1
The Luhmann-ID method described in #ZK-1a allows us
to maintain a free-branching structure:
ZK
├── ZK-1
│ ├── ZK-1a
│ ├── ZK-1b
| | └── ZK-1b1
| └── ZK-1c
| └── ZK-1c1
└── ZK-2
The simplicity of writing one idea per card is liberating; all I need to do is capture a thought and link it to a previous one, and later on I can trace the thread of thought to create even more complex ideas. Writing down individual ideas in this way and adding them to the collection of cards in a free-branching manner, it is always possible to recombine related cards in various permutations to give expression to new ideas.
There's more to Luhmann's method, and as I learned to use it and conversed with ChatGPT about it, I was struck by how much easier it was to work with GenAI when I started to share with it the items that are organised with my Zettelkasten's numbering system in my writing process. Is this how I can get my analog and digital second and third brains to communicate?
A Product Backlog is a shared brain for a Scrum team
Working with the index cards that comprise my Zettelkasten, I was brought back to memories of the round-table workshops that I had during the process of getting my CEO-founder a.k.a stakeholder to work with the team to generate our first Product Backlog for our Scrum team.
Before we did that, in the months before we got to work, the founder had written a long product requirements document. The problem with this method is that the product document was being rewritten every other week. Each version was meant to be clearer than the last, but after a while nobody could tell which ideas were current, which ones had been dropped, and which belonged to three drafts ago. The document had become a maze of revisions instead of a map. Any change we wanted to make had to be communicated through a series of comments and notifications.
In a Product Backlog, like a Zettel, the archetypal base unit is the user story that can fit into an index card. Over several sessions, I showed the team how to decompose that giant document into small user story cards around a table. I often emphasised that the product backlog is really just a set of hand-sized ideas that can be picked up, sorted, and revisited over time. Importantly, it really should not be hard to express what the user wants from the product.
When everything was laid out on the table, we could all see the complete picture of the product and a pretty good idea of how we each understood it, as we discussed who put each card there and why. Now the overall idea of the product was easily editable, each requirement precisely located. We could pick up any user story and discuss it, to decide to build it or not, or to throw it into the pile of "delightful" or "good-to-have". Just like our product, it was collaboratively assembled bit by bit, just small pieces loosely joined.
A well-maintained Product Backlog is what helps the Scrum Master and Scrum Team to organise their work well. The ideal of a great product backlog is its clarity and simplicity in communicating what the product is and will be, in a form that is minimal and loose enough to be useful as a daily tool by the whole team, shall I say like a shared brain for the team?
Anyone who has kept and nurtured any kind of documentation on a tool with a built-in GenAI assistant like Notion will quickly understand how powerful it is when you begin to accumulate atomic notes in your repository.
So that was how I started to develop my idea in this article, to try to use the Zettelkasten method and apply my learnings from collaborative product development to make a kind of "product backlog" so that I could make a shared brain between myself and my GenAI tool that would guide us when we are prototyping new products together.
Co-creating a product with GenAI with Zettelkasten
Teaching the robot Zettelkasten
Learning to make a Zettelkasten started a new daily writing habit for me. It was easy to learn how to do it from some videos, and simple to practice, because all that's needed is to write a few sentences on a small card every now and then. So I thought it might be fun to start my journey of creating my first product with AI-assistance, by first working with ChatGPT to make a Zettelkasten-based product backlog just for the two of us later, just like how I'd start a new project at work.
Here we are today: I'm pretending to my GenAI buddy that I’ve bought some index cards, Sharpies, and highlighters for it to use, and I have asked it to discuss my notes with me, to help me come up with some example texts for each card. You can follow along to see what we have made together from our discussions.
Let’s build it!
What you need:
- Index cards
- Sharpies / pens
- magic markers / highlighters
You can absolutely make your own digital Zettelkasten, especially to make it easier to manage and integrate with your AI tool. The point is, you don’t need a computer to maintain your product backlog, and it is important that you write your thoughts out yourself.
1. Describe your product in one sentence
First, let's write down the title and a short description
of your idea.
GENERAL-0
AI NOTE-TAKING APP FOR AI-WHISPERERS
Because someone has to keep track of your real thoughts.
2. Describe your product strategy
Now I’d like to define some basic ideas about the product. Let’s start with the age-old questions that define your product strategy: what is the purpose of this app, who is it for, and why they should use your product.
GENERAL-1
PURPOSE
A to-do app for AI-whisperers who drift across
multiple projects and need a place where their
ideas can form before they evaporate like breath.
GENERAL-2
WHO IS IT FOR
- Creators who converse with their models more than with other humans
- Solo builders running three prototypes, two side quests,
and one existential experiment at the same time
- Anyone whose daily workflow is basically a constellation of small tasks orbiting a very bright mind
GENERAL-3
WHY SHOULD THEY USE THE PRODUCT
- It keeps their scattered brilliance in one coherent timeline
- It shows what deserves attention next, even when every idea seems interesting
- It reduces context switching so their momentum doesn’t shatter mid-flow
GENERAL-4
HOW IS THIS PRODUCT DIFFERENT FROM OTHERS
- It assumes you’re running multiple AI projects because of course you are
- It stays lightweight so it never competes with your actual thinking
- It keeps tasks visible and grounded, letting your mind wander everywhere else it needs to go
3. Get your pitch ready
You’ll be making a lot of speeches and pitches, buddy. Let’s create a category of cards for your PITCHES and we create another card that combines the ideas of the first three cards that make it into your so-called Elevator Pitch.
PITCHES-1
ELEVATOR PITCH
A lightweight to-do app for AI-whisperers juggling
multiple projects. It keeps your ideas grounded,
shows what matters next, and stays out of your way
while you think.
Refs: #GENERAL-1, #GENERAL-2, #GENERAL-3, #GENERAL-4
Now, ask your AI buddy to generate a few more elevator pitches for you, according to the contents of the first 4 cards. I’m sure you know how to get it to sound just the way you want. Experiment with the variations until you find a version that feels natural for you.
4. Get organised with some product
Now we have some general cards, let’s lay out some basic categories that we might need to keep your cards organised. Here are some categories that we will be frequently thinking about as we learn about our users and iterate through our product:
GENERAL-5
CATEGORIES WITHIN THIS ZETTELKASTEN
- GENERAL: High-level notes about the product and its purpose
- PITCHES: Short descriptions that capture what the product is and why it matters
- PERSONAS: Portraits of our target users and their needs
- VALUE: Value propositions to our target users / customers
- ASSUMPTIONS: Assumptions about user behaviour that need to be validated
- FEATURES: Descriptions of features that deliver our value propositions
- JOURNEYS: Descriptions of user flows, collecting the features into coherent paths
- METRICS: Measurements that matter to track how our product is performing
Note: These aren’t prescriptive categories, and you will likely add more to suit your personal style, or not to use a prefix or category at all. I chose only a few categories to show how your Product Backlog will hang together. In general, prefer to have as few categories as possible or not at all.
Let’s add a couple of cards under GENERAL-5, that expands on what PERSONAS and VALUE prefix means.
GENERAL-5a
CATEGORY: PERSONAS
Cards in this category describe the people who
might use the product. Each card captures a simple
portrait of a user, their needs, and the problems
they want to solve.
REF: #GENERAL-2 defines the kinds of people we should
develop personas for.
GENERAL-5b
CATEGORY: VALUE
Cards in this category describe the value we
offer to our users. Each card focuses on one benefit
or outcome the user gains from the product.
REF: Our value propositions should be developed
based on the Value Proposition Canvas
https://www.interaction-design.org/literature/topics/value-proposition-canvas
And you can continue to generate the cards GENERAL-5c, -5d, etc. for each category.
5. Create some user personas
In GENERAL-2, we've defined our who our product is for, so we can create some user personas that allow us to visualise customers. Here I am introducing the idea of the Value Proposition Canvas, a modelling tool that I like to use to think about how my users will interact with my product. According to the Value Proposition Canvas, this is our starting point for creating a customer profile so we can define their “jobs” or outcomes they want, and their pain points or gain points.
Let’s start with one persona. Personas allow you to imagine a real person or user and analyse them. Think about the different kinds of users that you might encounter. Use the Value Proposition Canvas to generate a few more personas.
Value Proposition Canvas, Strategyzer
PERSONA-1
The Context-Switching Creator
A digital creator who moves rapidly between drafts,
models, prototypes, and publishing. They don’t feel
overwhelmed so much as constantly interrupted by
their own ideas. They want a tool that gives structure
without slowing their pace.
PERSONA-1a: CUSTOMER JOBS
- Capture ideas while working
- Track small tasks tied to different creative cycles
- Mark progress across projects
- Return to paused work without reorienting
- Keep experiments linked to their parent project
PERSONA-1b: GAINS
- A clear timeline for each creative thread
- Easy re-entry into a paused task
- A sense of continuity across experiments
- Reduced cognitive overhead
- A workspace that adapts to rapid shifts
PERSONA-1c: PAINS
- Losing momentum when switching tools
- Forgetting where they left off
- Scattered notes across platforms
- Disrupted flow during creative bursts
- Difficulty knowing which task belongs to which project
6. Design your value propositions
Each persona gives you ways to think about your Value Propositions. Using the Value Proposition Canvas, we capture these with the idea of the Product or Feature, and define the Pain Relievers and Gain Creators. We will make our first value proposition VALUE-1 based on PERSONA-1.
VALUE-1
Quick Capture
PRODUCT / FEATURE:
A fast way to record ideas instantly while staying in flow.
GAIN CREATOR:
Reduces cognitive load by offloading thoughts immediately.
PAIN RELIEVER:
Prevents disruptions when ideas arrive mid-task.
REFS:
- JOB: #PERSONA-1a “Capture ideas while working”
- GAIN: #PERSONA-1b “Reduced cognitive overhead”
- PAIN: #PERSONA-1c “Disrupted flow during creative bursts”
At the same time, let’s also design the Features that deliver our Value Proposition. For illustration, I am showing how two different features are related to one value proposition, and how user stories clarify how each feature will be seen from a user-centric view.
FEATURE-1
Inline Idea Capture
A lightweight input box available at any moment in the app so creators can jot down an idea without switching screens.
REFS:
- VALUE: #VALUE-1 “Quick Capture”
FEATURE-1a
USER STORY: Instant Capture in Flow
As a context-switching creator,
I want to capture an idea without leaving what I’m doing,
so that I can stay in flow.
REFS:
- PERSONA: #PERSONA-1
FEATURE-1b
USER STORY: One-Action Save
As a context-switching creator,
I want to save an idea with one action,
so that I don’t lose it while I’m thinking about something else.
REFS:
- PERSONA: #PERSONA-1
FEATURE-2
One-Tap Save
A single action that stores an idea instantly and links it to
the current project without requiring extra steps.
REFS:
- VALUE: #VALUE-1 “Quick Capture”
FEATURE-2a
USER STORY: Save Without Thinking
As a context-switching creator, I want to save
an idea with one tap, so that I can keep my
attention on the work in front of me.
REFS:
- PERSONA: #PERSONA-1
- FEATURE: #FEATURE-2
FEATURE-2b
USER STORY: Auto-Link to Project
As a context-switching creator, I want my
captured idea to attach itself to the right
project automatically, so that I don’t have
to organise anything in the moment.
REFS:
- PERSONA: #PERSONA-1
- FEATURE: #FEATURE-2
And finally we create a JOURNEY card that shows how the user would use the feature:
JOURNEY-1
Journey: Capturing an Idea in Flow
A brief path showing how a creator uses Quick Capture while working.
STEPS:
1. The creator is in the middle of a task on an AI-driven project.
2. A new idea appears mid-flow.
3. They open the Inline Idea Capture box without leaving the screen.
REF: #FEATURE-1
4. They type the idea in a few words.
5. They save it with a single tap.
REF: #FEATURE-2
6. The idea auto-links to the current project.
7. The creator returns to their work without losing momentum.
OUTCOME:
The creator keeps their pace, the idea is safely stored,
and there is no cognitive cost to switching modes.
REFS:
- VALUE: #VALUE-1 “Quick Capture”
- FEATURES: #FEATURE-1, #FEATURE-2
- PERSONA: #PERSONA-1
7. Keep generating personas and your value propositions
Repeat steps 5 - 6 until you have generated a set of personas, value propositions, features and journeys. It’s great to brainstorm with AI and other collaborators here, but you must always remember to check your assumptions, which leads us to the next point.
8. Check your assumptions
It’s important to realise that so far you probably have made a lot of assumptions about your users when you made the personas. You imagined their jobs, how they wanted to get it done, and what you think made it difficult for them to complete their jobs. Then you invented the features you wanted to make for them that you imagine would create a value proposition based on those assumptions. You have even imagined how they would use your features.
For example, ASSUMPTION-1, while being a safe assumption that people tend to be forgetful, when examined further with ASSUMPTION-1a also tells us that our product may not be as valuable if they have an easier and more immediate way of capturing ideas. Do they even need to capture so many thoughts?
So we need to keep track of the assumptions we've made, because assumptions are risks that must be validated.
ASSUMPTION-1
Creators lose ideas if they cannot capture them immediately.
REFS:
- VALUE: #VALUE-1 “Quick Capture”
- PERSONA: #PERSONA-1a “Capture ideas while working”
ASSUMPTION-1a
Creators do not already have a faster or easier
way to capture their ideas.
ASSUMPTION-1a1
EXPERIMENT: Test Current Capture Methods
Ask a small group of creators to show how they currently capture ideas.
Record how long each method takes and how often it interrupts their work.
If faster methods already exist, this assumption does not hold.
REFS:
- VALUE: #VALUE-1 “Quick Capture”
ASSUMPTION-1b
Creators want to capture all their ideas.
ASSUMPTION-1b1
EXPERIMENT: Idea Capture Diary
Ask creators to log every idea they have for one or two days.
Compare the number of ideas generated with the number they choose to capture.
If they capture only a fraction, this assumption needs revision.
REFS:
- ASSUMPTION: #ASSUMPTION-1b
- PERSONA: #PERSONA-1a “Capture ideas while working”
9. Define your measurements
Finally, we need to take some smart measurements to know if we are actually delivering the value that we are promising. These are the METRICS, of course. What could these cards look like?
METRIC-1
Time to Capture (TTC)
How long it takes a creator to record an idea from the moment it appears.
WHY IT MATTERS:
Measures whether Quick Capture actually keeps creators in flow.
TARGET (example): < 3 seconds
REFS:
- VALUE: #VALUE-1
- FEATURE: #FEATURE-1
METRIC-2
Capture Completion Rate (CCR)
The percentage of ideas that get successfully captured when the
creator attempts to note them down.
WHY IT MATTERS:
If creators still lose ideas, the feature isn’t doing its job.
TARGET (example):
> 90% of attempted captures saved
REFS:
- VALUE: #VALUE-1
- ASSUMPTION: #ASSUMPTION-1
10. Analyse and refine your ideas
Here is the round-up. By now you would have created a set of cards or text files that have the following contents. Over time, you will be adding and pruning branches, rewriting and merging cards.
ROOT
│
├── GENERAL
│ ├── GENERAL-0
│ ├── GENERAL-1
│ ├── GENERAL-2
│ ├── GENERAL-3
│ ├── GENERAL-4
│ └── GENERAL-5
│ ├── GENERAL-5a (PERSONAS)
│ └── GENERAL-5b (VALUE)
│
├── PITCHES
│ └── PITCHES-1
│
├── PERSONAS
│ └── PERSONA-1
│ ├── PERSONA-1a (Jobs)
│ ├── PERSONA-1b (Gains)
│ └── PERSONA-1c (Pains)
│
├── VALUE
│ └── VALUE-1 (Quick Capture)
│
├── FEATURES
│ ├── FEATURE-1 (Inline Idea Capture)
│ │ ├── FEATURE-1a (US: Instant Capture in Flow)
│ │ └── FEATURE-1b (US: One-Action Save)
│ │
│ └── FEATURE-2 (One-Tap Save)
│ ├── FEATURE-2a (US: Save Without Thinking)
│ └── FEATURE-2b (US: Auto-Link to Project)
│
├── JOURNEYS
│ └── JOURNEY-1 (Capturing an Idea in Flow)
│
├── ASSUMPTIONS
│ ├── ASSUMPTION-1
│ ├── ASSUMPTION-1a
│ │ └── ASSUMPTION-1a1 (Experiment)
│ └── ASSUMPTION-1b
│ └── ASSUMPTION-1b1 (Experiment)
│
└── METRICS
├── METRIC-1 (Time to Capture)
└── METRIC-2 (Capture Completion Rate)
Have you been discussing your notes with AI so far? How is it going for you? While your mileage may vary, I was quite pleased that by now my simple Zettelkasten product backlog seems to work very well with my ChatGPT. Without much sophistry, I was able to get neatly formatted and logical responses for prompts like, "come up with a couple of ideas for FEATURE-1 and FEATURE-2".
What's important is keeping your ideas about your product organised with this principle of atomicity and connection-building, as well as with the structure I have provided. Then just feed it all to GenAI to see it perform magic with its pattern-matching capabilities. Try these prompts with ChatGPT once you have put your backlog together, some questions that you might usually want to ask when grooming your Product Backlog.
PROMPT-1
Prompt for Reviewing Your Backlog
“Look at these cards as a whole.
What assumptions am I making, what risks do you see,
and which feature or value proposition depends on the
riskiest one?”
Why this works:
- Surfaces hidden assumptions
- Highlights the riskiest dependencies
- Helps you prioritise what to validate first
PROMPT-2
Meta-Prompt to Study Your Backlog
“Given my backlog, generate a list of questions I
should ask to interrogate the product: questions
about value, users, risks, features, gaps, contradictions,
and anything I may have overlooked.”
Why this works:
- Encourages deeper thinking through guided questioning
- Reveals blind spots you didn’t realise were there
- Shifts the AI from answering to helping you explore
- Produces a reusable set of prompts for future refinement
PROMPT-3
Prompt for Reducing the Product to Its Smallest Useful Form
“Given my backlog, identify what can be removed,
simplified, or delayed. Which features, stories,
or ideas are unnecessary for the core value,
and what is the smallest version of this product
that would still be meaningful to the user?”
Why this works:
- Forces clarity about what truly matters
- Reduces overbuilding and premature complexity
- Helps reveal the real Minimum Viable Product
- Keeps the backlog focused on delivering value, not volume
PROMPT-4
Prompt for Finding Contradictions and Inconsistent Thinking
“Scan my backlog for contradictions, mismatched
assumptions, or inconsistent logic. Identify
where personas, value propositions, features,
or journeys disagree with each other, or where
the product seems to make conflicting promises.
Show me every place where the thinking doesn’t line up.”
Why this works:
- Exposes contradictions that weaken product strategy
- Reveals where assumptions clash or features don’t match user needs
- Helps tighten the conceptual coherence of the product
- Prevents logical drift as the backlog grows
The Product Owner in a box
Exploring different permutations of Zettels with Grok
We’ve taken the first steps to assemble a structured set of notes that cover several core building blocks of product thinking — each expressed as small, atomic index cards you can shuffle, link, and recombine.
Let’s examine the features of the product backlog you now have:
- You defined the purpose of the product, who it is for, why it matters, and how it differs from other apps. These cards are the anchor for the entire backlog.
- You have at least one elevator pitch that will help you introduce the product to anybody.
- You defined some useful categories for your Zettelkasten. Although you might prefer different categories for yourself, I hope I have shown how each category functions in my system. What’s important: these categories aren’t hard rules; they’re just scaffolding to help your thinking grow.
- You built your first persona using the Value Proposition Canvas. This gives you a user-centric method for product development. When I practice this method, I find it easy to interview and discuss with anybody how they want to use my product.
- You wrote your first value proposition. This is important for understanding why someone would want to use — or pay for — your product.
- You designed two features that deliver your value proposition, each with its own user stories. You learned to write user stories that express, in the user’s own perspective, how they will use your product.
- You created a short journey showing how a real user would experience Quick Capture end-to-end. This ties the features together and tests whether they make sense in context.
- You documented the assumptions you unknowingly made and designed simple experiments to test them.
- You identified meaningful metrics that will provide measurements that matter.
- You have some ideas of how you can work with an AI companion to understand your product.
All that, with just a handful of cards! As your collection grows and you create more interlinked ideas, you will start to see how your product backlog becomes useful for generating new ideas that are focused on your users' needs.
I used the Zettelkasten method to structure this article as I wrote it with an AI companion.
While I have yet to try to make a product totally made with GenAI, I've discovered that I could at least start to develop my blueprint with it sanely, and that's a great first step. I was amazed by how much GenAI was able to contribute to the product backlog in this framework without it feeling like it was going out of control.
As I co-created this article with my GenAI buddy, I learned that it’s a tool that I can actually teach and make it work according to my processes, and Zettelkasten works not just for my memory — it becomes a way of showing the machine how I think, one card at a time.
Next: In the days leading up to Christmas 2015, read other articles from Advent of Tech @ Onepoint x Wepoint 2025.
References
I learned mostly from this guide to Zettelkasten to get started, https://zettelkasten.de/overview/
The Value Proposition Canvas is a technique popularised by Strategyzer AG. I use it often to think through the needs, pains, and gains behind every project I take on. Learn about it here: https://www.strategyzer.com/value-proposition
The Interaction Design Foundation is a great resource to learn user-centric design methods. Start with their page about the Value Proposition Canvas https://www.interaction-design.org/literature/topics/value-proposition-canvas?srsltid=AfmBOoq8JmqpwJv4QDfGA55wnB6FugYwtzyUDJa07AAav2FMOTa80FOp
I adapted my Zettelkasten numbering system from Kathleen Spracken, who teaches a minimalistic approach: https://www.youtube.com/watch?v=2a5TOzxuqxE
I first learned about Zettelkasten from my friend Liy. Knowing how much she writes every day, I was intrigued by the method that finally worked for her. Here is her experience: https://liyyusof.com/y5-zettelkasten/
Another set of lessons that helped me build my first Notion-based Zettelkasten came from Herbert Lui: https://herbertlui.net/8-lessons-from-800-note-cards-in-the-zettelkasten/
Top comments (0)