The story
Copilot Mobile was becoming more capable, but the Create tab still carried
legacy behaviours. It behaved like a collection of tools: users had to
decide what to make, choose the right entry point, understand which flow
belonged where, and then navigate through different creation patterns.
That created too much decision-making before creation even started.
Create held scan flows, document flows, image and poster flows, text
drafting, PDF paths, templates, and emerging AI artefacts. Each had value,
but the risk was that users would meet Create as a scattered set of options
instead of a clear journey. The design challenge became one question:
How do we make Create feel like one coherent Copilot experience, not a
folder of unrelated tools?
Why this mattered
Mobile creation has different constraints than desktop. On desktop, people
often arrive with a file format in mind. On mobile, they arrive from a
moment: a photo, a note, a thought, a scan, a quick draft, a poster idea,
something they need to turn into a usable artefact fast.
So Create had to support two behaviours at once. Fast start,
where users don't need to understand the whole product map before creating.
And flexible continuation, where once an artefact exists
they can edit, refine, share, recall, or continue from it. That's why the
redesign needed not just a new homepage, but prompt patterns, templates,
bottom sheets, result actions, loading states, recall surfaces, and
accessibility built into handoff.
What we designed
An opinionated Create homepage and IA
The homepage was rebuilt to be more opinionated and easier to enter,
reducing decisions and pulling users into creation quickly. The new
architecture grouped Create around practical jobs, capture, draft, visual
creation, and documents, and left room to scale as new artefact types
arrived.
Prompt-led creation
Prompting became central, with consistent copy and behaviour across text,
poster, and image, plus a suggestion-pill pattern. Instead of choosing a
tool first, users could express intent first.
Capture as a first-class input
Capture was treated as a core mobile-native mode: scanning, text from
image, translate, image to document, and more. It turns the camera into a
creation input, not just a utility, and gives Copilot real-world context.
Visual, document, and result flows
Image and poster creation gained bottom sheets for controls, style and size
attributes, and clear loading and result states. Document flows stayed
available without letting Create become a document launcher. Result screens
became continuation surfaces, with a focused "Ask Copilot to edit" control
that guides people toward refinement rather than an open-ended prompt.
Async, accessibility, and platform states
Longer-running artefacts needed background execution, progress messaging,
and notification patterns so people could leave and return without losing
work. Accessibility annotations, screen-reader labels, focus order, and
button labelling, were built into Figma handoff. And the system accounted
for tablet, landscape, and plan-specific and ineligible states.
The turning point
The turning point was moving from tool-first IA to intent-first creation.
Before, Create risked becoming a list of things the app could do. After,
it became a guided system for moving from intent to artefact.
Capture became an input system, not just scan. Prompting became the centre
of creation, not a secondary field. Templates became guidance, not the
entire model. Result screens became continuation surfaces, not endpoints.
Recall became part of the loop. And accessibility became part of handoff,
not a final QA pass.
Design principles
Start with intent, not format
Users shouldn't decide whether an idea is a poster, document, PDF, or image before they begin. Create should translate intent into the right artefact.
Reduce decisions before creation
The fewer choices someone makes before starting, the faster they get to a result.
Make mobile creation feel native
Prioritise capture, bite-sized flows, and quick journeys built for a phone.
Design for continuation
Editing, refining, sharing, and recall are part of the creation journey, not afterthoughts.
Build for many artefacts, not one launch
Patterns had to scale to documents, visuals, and future Copilot artefacts, so nothing could be bespoke per type.
Design decisions
Rebuilt the Create landing page
A more opinionated, AI-forward homepage that reduces the number of choices
before starting.
Prioritised capture and AI artefacts over documents
Surfaced quick, mobile-first creation more prominently than document
creation, which benefits from a phone's strengths.
Made prompting consistent across artefacts
One mental model for the prompt box across text, poster, and image, with
structured suggestions to reduce blank-page friction.
Defined result-page editing and bottom sheets
A clear "Ask Copilot to edit" control with proper states, and secondary
controls (settings, image options, refiners) moved into bottom sheets so
the main canvas stayed focused.
Designed for async, accessibility, and plan variants
Progress and notification patterns for long-running generation,
accessibility annotations in handoff, and variants across free, paid,
enterprise, and ineligible states.
What we learned
The biggest learning was that mobile creation needs stronger guidance than
desktop. People don't arrive knowing whether they need a document, poster,
PDF, scan, prompt, or template. They arrive with a job to do.
Research reinforced it: the Create tab was discoverable, but people could
get sidetracked inside it, and the word "Create" was strongly linked to
"document," leading some users into the wrong section. Once they entered
the intended flows, tasks like converting handwritten notes and creating a
poster rated well for ease of use. So the priority shifted from exposing
every capability equally to guiding people into the right path with fewer
wrong turns.
Outcomes
The redesign helped Create become a recognised AI-forward creation
experience inside Copilot Mobile. A new Create tab focused on Copilot-based
creations shipped to internal channels, followed by document workflows with
templates and a "My Creations" page for recall.
Beyond adoption, the strongest outcome was the breadth of systemisation: a
clearer AI-first landing experience, a scalable IA for future artefacts,
prompt-led creation across text, image, poster, and documents,
capture-first workflows, result-page edit and continuation patterns, async
and notification thinking, accessibility built into handoff, plan-aware
variants, and recall that brings creation back into the broader Copilot
loop.
Looking back
The biggest lesson was that creation isn't a single flow. It's a loop:
start, guide, generate, refine, save, recall, continue. The redesign was
valuable because it reworked that whole loop, not just the screen, moving
Copilot Mobile from a place to find creation tools to a place to start with
intent and let Copilot shape the output.
What's next
The next opportunity is to evolve Create from an AI-first tab into a full
creation system across Copilot: richer prompt templates, better recall,
stronger chat-like refinement, scalable async generation, clearer
plan-aware experiences, and future artefacts like charts, podcasts,
notebooks, and agents.
Create should become the place where Copilot turns intent into artefact,
whether you start with a photo, a thought, a file, a prompt, or a template.