Skip to content
CV
WorkUWX · Unified Work Experience13 min read01 · Overview

Initiative case · AI experience · Systems

A card for sensitive data became a work model for every product.

UWX, the Unified Work Experience, describes how people and AI work together across the zvoove products: from the signal to the result. It started with a GDPR card in zAIn, which became a pattern and then a model. Today product managers build their prototypes with it, and UWX is anchored in the 2026/27 product strategy.

Read the caseRJump to results
Role in the project
Author of the modelDeveloped UWX on my own
Setting
zvoove GermanyOnly designer, with product management and engineering
Responsibility
From the card to the modelGDPR card, three-layer pattern, UWX, PM guide, vision-system prototype
Status
Established+3

As of September 2026

  • Established

    In use: UWX with the product managers and their teams, anchored in the 2026/27 product strategy

  • Live

    GDPR cards in zAIn, reading and planning, never writing

  • Prototype

    Vision-system prototype (3/6/9), September 2026, used by PMs; 3/6/9 remains a framework

  • Open

    There are no impact figures for UWX

and 3 more states
As of September 2026
Established+3

As of September 2026

  • Established

    In use: UWX with the product managers and their teams, anchored in the 2026/27 product strategy

  • Live

    GDPR cards in zAIn, reading and planning, never writing

  • Prototype

    Vision-system prototype (3/6/9), September 2026, used by PMs; 3/6/9 remains a framework

  • Open

    There are no impact figures for UWX

I did not build another product. I built a model for how people and AI work together across every product.

UWX · Unified Work Experience

At a glance

  • 3 layers

    understand, plan, result

    3 layersunderstand, plan, resultThe three-layer pattern from the GDPR card in zAIn
  • 2026/27

    anchored in the product strategy

    2026/27anchored in the product strategyEstablished with the product managers and their teams
  • 3/6/9

    vision-system prototype, used by PMs

    3/6/9vision-system prototype, used by PMsSeptember 2026; 3/6/9 remains a framework
  1. Problem

    One approval does not cover everything

    The GDPR card became the three-layer pattern: understand, plan, result.

  2. Intervention

    A model instead of another product

    UWX sets how a case runs from signal to result and when a person decides.

  3. Outcome

    The model lives in the PMs’ tool

    In the 3/6/9 prototype UWX is the operating system; the 2026/27 strategy builds on it.

Role in the project
Author of the modelDeveloped UWX on my own
Setting
zvoove GermanyOnly designer, with product management and engineering
Responsibility
From the card to the modelGDPR card, three-layer pattern, UWX, PM guide, vision-system prototype
02 · The card

It started with a card for sensitive data.

zAIn reads operational data straight from the ERP, including data about employees. As soon as a question touches sensitive fields, zAIn has to ask first, and it has to ask in the conversation itself. For that I designed the GDPR card: a confirmation in the chat before zAIn accesses sensitive employee data.

Illustration, rebuilt. The GDPR card in the conversation: zAIn asks before it reads sensitive employee data. Sample data.

The card is one of the confirmation and reflection cards I designed for zAIn. zAIn has been live since 10.08.2026. The GDPR cards went live as reading and planning cards, not as writing ones.

The card showed that a single confirmation is too coarse. Approving a flow and approving a protected value are two different questions. That is where the pattern in the next chapter came from.

03 · The pattern

The card became a pattern with three clear jobs.

For sensitive data in zAIn I built the three-layer pattern, in the GDPR context and by GDPR rules. It grew out of two questions: how does zAIn handle sensitive data, and what changes when it does not just read but writes?

Three-layer patternAnatomy of one answer · example
  1. Message · What did zAIn understand?

    Interpretation, questions back, handover. Nothing here can be acted on.

    Jonas Weber is off from Monday. Can you find a replacement for his assignment?A
    Understood: a replacement for Jonas Weber from Monday, assignment EIN-1042 at Nordhafen Logistik.
  2. Intent · What is zAIn about to do?

    The steps in the order they will run, in the words dispatchers use. Until the plan is approved, nothing runs.

    Plan · 4 stepsApproval required

    Look up Jonas Weber → find assignments → read requirements → match who is available

    Runs after approvalDeclineApprove
  3. Result · What came back, and what now?

    Result, selection, draft. Only here may the flow stop and ask consent for a protected field.

    Three employees fit. Whom should I ask?

    3 matching employees

    Anna BergerForklift licence · 12 kmfrom Mon
    Mara JensenForklift licence · 8 kminduction
    Tom KrügerForklift licence · 21 kmfrom Wed

    Sample data

Rebuilt from my pattern documentation. The simple case, cover for a shift, through all three layers; examples paraphrased.

Each layer answers exactly one question, which is why each can also stand alone. A plain answer needs no plan approval; a notification the system raises on its own is only a result.

Six shapes, one grammar

Shape01Message02Intent03ResultTypical trigger
Messagewithout Intentwithout Result“When does the early shift in Kiel start?”
without Messagewithout IntentResultfrom the systemForklift licence expires in 14 days
MessageIntentwithout Result“Put Anna Berger on the early shift.”
Messagewithout IntentResult“Who is free next week?”
without MessageIntentResultscheduledFriday 2 pm: missing timesheets
MessageIntentResult“Can you find cover for Jonas Weber?”
Rebuilt from my pattern documentation. Layers are added by need, never by default.
zAIn as a full page: a question about the key figures, answered with a table and a short note
zAIn, live: a reading question needs no plan and no approval. The answer comes as text with a table. Synthetic data.

Not every field deserves an interruption.

What earns a consent

L1 · operational

Read freely

What dispatchers already see. The plan approval covers it.

Names and employee numbers
Roles and qualifications
Availability and location
L2 · personal

Read, but logged

Shown without interruption; the access goes into the log.

Private address and phone number
Contract terms
Absence reason, category only
L3 · protected

Masked until consent

Never shown in full without consent for exactly this field. The plan approval does not cover it.

Bank details
Social security number
Health records
Illustration, rebuilt. The level belongs to the field and is set before the plan is shown. Sample data.

Salary, invoices, bank details and sickness reasons cannot be reached in zAIn v1; they are not just hidden. So the pattern also describes the path for when that boundary opens. At GA zAIn reads; write actions run only on test and staging data.

Two requests, one structure

Jonas Weber is out until Friday. Can you find cover for his early shift?
I can take that. Before I start, this is how I would go about it:
To try out, rebuilt from my pattern documentation. The first request reads assignment data and ends in a message; the second stops at the protected field and asks, then again before writing. In zAIn v1 bank details are not reachable. Sample data.
04 · The model

What worked in one chat had to hold across every product.

The pattern settled the question for one conversation. At zvoove, though, work runs across many products: Launchpad, Notification Center, zAIn and the domain products. With UEE, the Unified Entry Experience, I had defined the shared entry and prototyped it in code; it became the Launchpad. What remained open was how a case moves across product boundaries without losing its context.

For that I developed UWX, on my own. UWX is the shared model for AI-supported work across the zvoove products and a design principle for Product, Design and Engineering. I set up the concept and the PM guide in Confluence in August 2026.

The life cycle of a caseeight stations in four stretches
  1. Signal

    Detect and sort

    What comes up becomes a case instead of vanishing in a list.

    • 01 Detect
    • 02 Prioritise
  2. Understand

    Understand and decide

    zAIn condenses context, cause and risk. A person chooses or confirms the next step.

    • 03 Understand
    • 04 Decide
  3. Get it done

    Act and close

    The domain product or an automation carries it out; handovers stay visible.

    • 05 Act
    • 06 Continue
    • 07 Resolve
  4. Value

    Make it visible

    What the case achieved shows at the end.

    • 08 Value

The state is the same in Launchpad, Notification Center, zAIn and the domain product. That holds for every station; it is not a ninth one.

  1. Signal

  2. Understand

  3. Decide

  4. Impact

Illustration, rebuilt from my UWX documentation. One case through the stations, from the signal in Launchpad to the result. Sample data.
Notification Center with an overdue shift; on the right zAIn suggests three matching employees
Live: a case from the Notification Center goes to zAIn with its context. One stretch of the life cycle, signal to understanding, not UWX as a whole. Synthetic data.

From the pattern I took the principle, not the structure: each level answers one question, and approval sits where a decision has consequences. The three layers remain the pattern for a conversation in zAIn. The levels L1 to L3 in the conversation became the autonomy levels of UWX.

How AI may helpautonomy in four levels
  1. Level 1

    Manual

    A person sees the case and does everything themselves.

    • needs a clear display
  2. Level 2

    AI assists

    zAIn explains the case and suggests something. A person decides.

    • needs sources and reasoning
  3. Level 3

    AI prepares

    The action is fully prepared and runs only after approval.

    • needs a preview and a log
  4. Level 4

    Autonomous

    AI handles simple cases itself, only within clear limits.

    • needs clear limits and a way back
    target

A case moves up a level only when context, permission, safeguards and confidence are ready. The person stays responsible.

My principle throughout: make it visible before making it conceptually perfect. I wrote it down in August 2026, right after the readiness audit.

05 · PM enablement

Product managers get the model whenever they prototype.

A model in Confluence helps little if every team has to translate it on its own. So UWX sits where new flows are made: in the vision-system prototype I built for the product managers in September 2026. With it PMs build live-like prototypes, and UWX is their operating system: anyone designing a new flow gets the model automatically. That is also where I finished defining UWX.

The vision-system prototype is the 3/6/9 prototype. 3/6/9 is my framework for readiness, prioritisation and prototypes; it is a way of working, not a product, and it does not ship.

Readiness first, then the prototype

Five product areas were at very different levels of maturity. In the readiness audit I laid them side by side on eight dimensions, with the perspectives of six product managers. From that I decided which flows got a deep prototype first.

Readiness audit

  • 5

    product areas

  • 8

    readiness dimensions

  • 6

    PM perspectives

The rules came first

The enablement started earlier. When everyone began working with AI assistants, I drafted the AI Playbook: for enablement, to sharpen what gets built, and to bring people back down to earth, because many feared for their jobs. Back then everyone used it. From 3 to 10 February 2026 I also built the AI Assistant Handbook on my own in Figma Make; the AI Product Playbook grew out of it. More in the zAIn case.

In July 2026 the AI Playbook 2.0 Lite for product managers followed: how to build with Claude, Claude Design and Claude Code, how to plug in skills from the zvoove Skills plugin and what to use for what, with weekly learnings and shared responsibility. In depth in the Skills case.

06 · Product strategy

A concept that was long held back now carries the strategy.

I wrote the master document for UEE in January 2026. Internally the concept was held back for a long time and not taken seriously.

I did not wait. I gathered evidence, internally and with customers, aligned with the VP of Engineering and the head of AI of another unit, explained the concept and kept pushing it. In the end the 2026 and 2027 product strategy was aligned to it.

UWX builds on that entry and is anchored in the same product strategy. In it, UEE is the layer for entry and orientation.

  1. GDPR card

    A confirmation in the zAIn chat before sensitive employee data is read. Live, reading and planning.

  2. Three-layer pattern

    Understand, plan, result; field consent kept apart from plan approval.

  3. UWX

    One model for AI-supported work across all products; concept and PM guide in August 2026.

  4. PM enablement

    The vision-system prototype (3/6/9) for the product managers, September 2026.

  5. Product strategy

    UEE and UWX are anchored in the 2026 and 2027 product strategy.

07 · Outcome

What is established and what remains open.

UWX is established and in use: the product managers and their teams work with it, and it is anchored in the 2026/27 product strategy. The parts underneath are at different stages.

In productionLive2 entries

  1. GDPR cards in zAIn

    Reading and planning, never writing.

  2. Launchpad, zAIn, Notification Center

    Parts that show pieces of UWX; live since August 2026, rollout ongoing.

Design & proof2 entries

  1. Vision-system prototype (3/6/9)

    Prototype

    Where UWX is fully defined and serves as the operating system. 3/6/9 is a framework and does not ship.

  2. Three-layer pattern

    Pattern

    For sensitive data in zAIn, by GDPR rules. The writing path is described; writing runs only on test and staging data.

Pending1 entry

  1. Impact

    Open

    There are no impact figures for UWX; baselines are missing.

Fewer context switches across products is defined as the target signal; it has not been measured. That parts are live does not prove UWX is implemented everywhere.

Learnings

  1. Start small, then generalise.

    The card was concrete enough to show what a general model has to solve.

  2. Evidence carries further than slides.

    The concept got through on the evidence I gathered.

  3. A model needs a tool.

    UWX only became usable in the PMs’ prototype.

What I would do differently today

  1. Today I would collect baselines early.

    Without starting values, the impact of UWX stays a claim.

A model takes hold when it lives in the tool others already work with.

fin (for now)

CV · Maximilian Dahl
In plain words

What do you actually do, Max?

My parents and family have nothing to do with tech or my profession. They told me they did not understand my website. This is for them. And for anyone who would like an explanation without the jargon.

I make software easier to work with.

When someone spends ages searching for a button or has to enter the same information three times, I want to know why. I watch how people work, talk to them and design a better way through the task. Then we try it together.

My title is Principal Designer. That means I also make decisions about how whole products fit together and help the team build on them. I still design and write code myself.

Here is a simple example.

At zvoove, I designed an assistant called zAIn. This simplified example shows the idea:

  1. Someone asks

    “How much holiday do I have left?”

  2. The assistant looks it up

    It finds the information and explains the answer.

  3. The person decides

    If they want to book time off, they review the details before anything is sent.

My job is to work out what people need to see, how the answer should be explained and when the program needs their approval.

A design system is a bit like LEGO.

Imagine every team inventing its own bricks. Nothing fits. With a shared kit, the pieces connect, even when people build different things.

Shared pieces. Different things you can build.
Book time offReview an invoice

In software, those pieces are things like buttons, input fields and messages, plus rules for how they work together. I built this kind of kit at Container xChange. It is called Compass.

What has come out of my work?

One starting point
At zvoove, I designed the shared starting point for several programs and tried it out in code. It became Launchpad, which has been live since August 2026.
Built, too
I built the visible part of the notification centre myself. A backend engineer built the part behind it, and we shipped it together.
A team that can build on the work
At Container xChange, I grew the design team from 4 to 14 together with Niclas. Compass gave designers and developers a shared kit to work with.

What am I particularly good at?

I look beyond the screen in front of me. When several teams run into the same problem, I look for a solution they can share. That is how Compass came about.

And I like making an idea tangible early. I can build a working version myself, put it in front of someone and find out where it falls short. I would rather learn that from using it than from a long discussion about a picture.

What would I like to do next?

I want to help decide what a product should become, support a team and keep building myself. I am looking for a role as a Principal Designer, Team Lead or AI Product Builder, in a team that builds with artificial intelligence.

In plain words: I want to take responsibility for the whole thing, without giving up the part I enjoy most — trying things out and making them work.