The brief records what needs to change and who it may affect. It remains a visible source, not a vanished first prompt.
A chat forgets.
A canvas remembers.
AI made drafts faster, but it left the design process scattered across chats and tools. Plooi asks what happens when the work itself lives on a canvas we can read, challenge and change.
Plooi
01 / THE PROBLEM
Different work.
Different work.
Same chat box.
In my day-to-day design work, I increasingly felt like a human copy-and-paste machine between AI chats and my design tools. The tools produced material quickly; I had to move context, reconstruct decisions and turn replies into usable artefacts. It raised a more fundamental question: how do we want to work as designers when AI is part of the team?
ONE BRIEF / THREE DESIGN METHODS
CHAT-ONLY
The format stays the same.
Design assistant ONE THREAD
A new task becomes another message. The result is still a block of text in the thread.
METHOD-SHAPED
The workspace changes with the task.
PERSONA / WORKING ARTEFACT01 / 03
Sam, frequent commuterRegional rail · illustrative draft
NEEDS
A valid ticket ready when connectivity drops.
FRICTIONS
Unclear updates during a tight transfer.
BEHAVIOUR
Checks the next connection before boarding.
OPEN QUESTION
Which information earns their trust?
Structured fields · Editable claims · Source context
JOURNEY MAP / WORKING ARTEFACT02 / 03
A trip unfolds across stages.
01PlanFind a route
02BuyKeep a ticket
03TransferCheck updates
04ArriveClose the trip
Stages · Moments of friction · Opportunities
INTERVIEW GUIDE / WORKING ARTEFACT03 / 03
A conversation needs a usable guide.
01
Tell me about your last transfer.
OPENING02
When did you check your ticket?
BEHAVIOUR03
Where did information fail you?
PROBEQuestion order · Prompts · Room to adapt
A persona is edited as fields. A journey is read across stages. An interview needs questions you can use in the room.
Illustrative comparison of interaction patterns. The next chapter shows Plooi’s actual interface.
Plooi’s response: make the method the interface, and keep the work connected on a canvas.
02 / ONE WORKFLOWFrom question to artefact.
From question to artefact.
All on one canvas.
Here is one product demonstration: a regional rail brief develops into a research workflow. Each method receives context from the previous one. The canvas lets the team return to any step without reconstructing the whole conversation.
Target group and interview guide make the next inquiry explicit. A synthetic interview can rehearse the flow; it is clearly marked as a simulation.
Patterns, contradictions and opportunities sit beside their sources. The team can challenge an interpretation instead of accepting a polished paragraph.
The persona becomes a working object with needs, behaviours and frustrations that can be edited and refined — and still points back through the chain.
Project brief01 / 04
The interview in this example is synthetic. It demonstrates the workflow; it is not presented as field research.
THE METHOD LIBRARY
One canvas for the whole design journey.
The rail example follows a few research methods. Across the wider library, a brief or first idea enters the Double Diamond at Discover. Methods help open up the problem, frame it, explore solutions and shape an outcome.
METHODS ACROSS THE DOUBLE DIAMONDExplore the full method library ↗
STARTBrief
or idea
or idea
01 / DIVERGEDiscoverInterview guide
Desk research
Desk research
02 / CONVERGEDefineInsight synthesis
How might we
How might we
03 / DIVERGEDevelopIdeation
Wireframes
Wireframes
04 / CONVERGEDeliverStoryboard
Product backlog
Product backlog
03 / INSIDE A METHODA method is more
A method is more
than a prompt.
A method needs an input, a recognizable form and a place for the designer to intervene. The overview stays compact on the canvas; the full artefact opens when the work calls for detail.
SYNTHESIS / REAL INTERFACE
A finding should carry its sources.
The Insights method separates patterns, contradictions and opportunities. Source labels remain attached to each finding, so the next decision has something to stand on.
S M L
One artefact, three reading distances. A glanceable card on the canvas, a working panel for editing, and an expanded view for the full method. The structure stays recognizable as the task changes scale.
04 / HUMAN CONTROLGenerate.
Generate.
Judge.
Keep what matters.
AI can propose a first pass. The designer can edit, add, validate or hold individual parts. When a method runs again, the curated work should not silently disappear.
That decision lives in the interaction and the data model: every item can carry its origin, review state and whether it should survive a re-run.
Explore the product documentation ↗INTERACTION MODELRegional rail · demonstration
INPUT
Travellers need reliable access to tickets and journey updates while moving between stations.
INSIGHT 01AI suggestion
Offline ticket access could reduce friction when connectivity fails.
INSIGHT 02AI suggestion
A faster route search may matter most during transfers.
Try keeping the first observation, then re-run.
Illustrative interaction based on Plooi’s method behaviour. No model is called here.
05 / DESIGN ITERATIONThe canvas was there
The canvas was there
from day one.
I started with a prototype in Google AI Studio. The canvas was already the central idea; the continuing design work has been about making every method, connection and change immediately legible. These are the decisions that shaped that iteration.
The first version already connected design methods on a canvas. Later iterations focused on making the workflow readable, editable and dependable.
View the original capture ↗Keep the project visible.
The brief, methods, outputs and their relationships need to be readable at once. A designer should see the state of the work without rebuilding it from a thread.
Give every method its own shape.
A journey is not a SWOT matrix; an interview guide is not a persona. Shared controls keep the system coherent, while each artefact gets a form appropriate to the method and the level of detail needed.
Make provenance operable.
Source, model output and human edits need distinct states. A visible trace lets someone inspect the basis of a claim; it does not pretend to expose the model’s inner reasoning.
Show when the work has changed.
If an upstream source changes, downstream artefacts may need another look. Freshness, versions and re-run controls turn that dependency into a decision the team can see.
Iteration happened continuously across interface and system behaviour. The instruction, parser, saved output, version and export all had to agree on what a method actually does.
06 / WHERE IT STANDSThe product works.
The product works.
Now it needs real users.
Plooi is a functional alpha. The canvas and method workflows can be used today. Several people have offered to test it, but there is not yet enough sustained use to claim an outcome for teams. Finding the right users and learning from their work is the next challenge.
BUILTA functional alpha
The core product exists and is available to explore.
LEARNINGEarly tester interest
Real usage and longer-term value still need to be observed.
NEXTReach & onboarding
Bring designers into the product, watch where the workflow helps and where it breaks.
THE IDEA IN MOTION
A concept film for a different way of working.
The film follows the move from scattered chat outputs to a connected canvas, where methods and their results stay visible. The interactive examples above show the product and its design decisions.
FROM A QUESTION TO A PRODUCT
How do we want to work as designers?
Plooi is my answer in progress: a shared canvas where AI can do the legwork and people keep the context, judgment and ability to change direction. The next iteration will be shaped by use, not only by what I can build.