← Home

Day One is a chat platform plugin for the moment someone joins a group they don’t know yet. The hard part of arriving isn’t the work, it’s the unspoken rules — when to post, how formal to be, whether a question is welcome — so newcomers stay silent and read for weeks. Day One reads the group’s own chat history, turns it into a short private guide, and drafts a first message that fits. Halfway through the project the research told us we were solving the wrong problem, and we let it change our minds.

IXD CAPSTONE · SUMMER 2026Team Project

Chat platform plugin

Day One

Reads the group. Guides the newcomer. Drafts the first message.

ROLE
Research · UX
UI · Prototyping
TIMELINE
14 weeks
TOOLS
Figma · design
Claude · synthesis
Perplexity · research
The Day One panel open beside a Slack channel, showing what the group talks about and the words it uses

A LOOK AT WHAT WE BUILT

The Day One listing in the Slack App Directory
Discovered in the Slack Marketplace
The private guide panel, built from the group's history
A private guide, built from the group’s history
Editable message starters sitting above the composer
Help writing the very first message

01THE PROBLEM

The hardest part of joining isn’t the work. It’s not knowing the unspoken rules, and being afraid to ask.

Unspoken rules Fear of asking Weeks of silence

They don’t know the norms. When to post, how formal to be, whether a question is welcome. So they stay silent and read for weeks. The barrier isn’t skill. It’s the risk of getting it wrong in front of everyone.

She lent me an eraser once when I’d forgotten mine, and we became friends. Early on it made me feel safe, knowing I wasn’t the only one in that situation.

Interview participant, on switching schools

And it is measurable across onboarding, peer support, and repeated exposure.

12%

of employees strongly agree their organization onboarded them well.

Gallup · via StrongDM

29%

drop in anxiety symptoms within 90 days when newcomers had peer support.

Peer-support study · PMC (NIH)

208

experiments confirm repeated exposure reliably builds confidence in unfamiliar settings.

Meta-analysis · 208 experiments

02THE SOLUTION

The group already explains itself. Day One just reads it back.

It reads the group’s own chat history and turns it into a short, private guide, then helps write a first message that fits.

01

Reads the history

Last 60 days, channels the host allows. Patterns only, nobody quoted by name.

02

Builds a private guide

Each newcomer’s own view: what the group talks about, when people post, who to ask.

03

Drafts the first message

An editable opener, so saying hello isn’t a decision made alone.

HOST · ONCE

Sets it up in ~2 minutes

Picks channels, edits the auto-drafted welcome, sets who counts as new. Then never touches it again.

one setup,
every newcomer

NEWCOMER · EVERY TIME

Gets a private guide, automatically

The moment they join: their own read-only guide and message help. Private, so nobody sees what they read.

03HOW WE GOT THERE · RESEARCH

We started wide: who feels new, and why?

Before building anything, we interviewed very different kinds of “new”: new hires, movers, career-switchers, looking for the pattern underneath.

Qualitative interviews

Across relocation, new jobs, career switches, and new personal practices.

4 participant types

From “new to a community” to “an insider who feels new,” mapped to compare.

Affinity mapping

Clustered every quote into themes, then pulled the recurring tensions out.

Affinity maps: clustering interview quotes into themes

Affinity board: interview quotes clustered into coloured theme columns
Affinity board: a second cluster of quotes grouped by theme
Affinity board: a third cluster of quotes grouped by theme

Six themes surfaced. Read together, they pointed one direction:

Unclear expectations Cultural gaps Feeling alone & unsure Peer support changes the outcome Repeated exposure builds confidence Small acts of participation

04THE PIVOT

Halfway through, the research told us we were solving the wrong problem.

Wrong problem Silent Researcher New direction

Our first instinct was loneliness. But the “Silent Researcher” persona showed the real barrier. It wasn’t being alone. It was not knowing how the group works, and being afraid to reveal it.

FIRST HMW

How might we help a newcomer quickly find someone in the same situation, so they feel “I’m not the only one”?

WHERE WE LANDED

How might we give a newcomer early, low-risk guidance into a group’s unspoken norms, so they feel “I know how this works”?

That shift changed the product itself, not just the words on a slide:

THE NAME

First Step GuideDay One

“First Step Guide” described a feature. “Day One” names the feeling: the first day, made survivable.

THE STRUCTURE

New channelRead-only panel

The early version made a public channel in everyone’s sidebar. Now: one private box, only the newcomer sees it, read-only.

05TEST BEFORE BUILDING

Before designing a single real screen, we tested the flow on paper.

We tested the flow on paper before investing in real screens. Three people, four tasks, host and newcomer. We watched for hesitation, skips, and whether they reached a first message.

PARTICIPANTS

Jack (IxD) · Seiya (Film) · Alex (Transportation)

TASKS

Download the agent · Personalize settings · Read group info · Write & send first message

METHOD

Paper prototype, run as a think-aloud walkthrough

Three frictions surfaced. Each one became a fix we carried into the real build:

WE SAW
People couldn’t find where to start. The agent’s entry point wasn’t obvious on first landing.
WE CHANGED
Made the Day One box the one mint-accented element in the sidebar, so it’s the first thing that stands out.
WE SAW
Setup showed too many options at once; users skipped past the summary preview.
WE CHANGED
Broke setup into numbered steps with a live preview of the newcomer’s view.
WE SAW
No path existed if a host declined to let the agent read history. The flow dead-ended.
WE CHANGED
Added a consent screen with an explicit, handled decline path.

With the flow validated, we built it for real. ↓

06THE BUILD

From validated flow to a real product.

The Day One panel lit against a dimmed Slack workspace, with a note reading: the lit panel is our plugin, everything dimmed behind it is the group's existing Slack

Host · sets it up once

The Day One listing in Slack's App Directory
01Install from the MarketplaceHost finds Day One in Slack’s App Directory and adds it to the workspace.
A consent screen listing exactly what Day One reads
02Grant access, transparentlyA consent screen spells out exactly what Day One reads, and what it never does.
The host editing an AI-drafted welcome message
03Edit the AI-drafted welcomeDay One drafts a welcome from the group’s own history. The host tweaks and publishes.

Newcomer · the first ten minutes

The guide panel ranking what the group talks about
04See what the group is aboutTopics ranked by how often they come up, plus the group’s own in-words a newcomer wouldn’t know yet.
The guide panel showing activity rhythm and typical reply time
05Learn when & how people talkActivity rhythm and reply time, so a slow reply reads as the group’s pace — not a snub.
Editable message starters sitting directly above the Slack composer
06Get help with the first messageEditable starters in the group’s tone, right above the composer. Nothing sends until they press send.

07DESIGN DECISIONS

Every screen came down to one principle: lower the risk of being seen.

LANGUAGE

We never say “anxiety” in the UI

The research is about anxiety; the product isn’t. In-app copy stays functional, like “when people talk” and “who to ask,” because naming the fear on screen makes the user feel watched.

PRIVACY

Anonymous by construction

Hosts control the guide with rules, not names: a join date, a message count. Patterns only, never who posts most or who read what. Privacy is the data model, not a setting.

PRESENCE

One quiet box, never a channel

The old version added a channel to everyone’s sidebar. Now: one dashed box only the newcomer sees. Dashed borders mark every surface the app owns.

RESTRAINT

Mint appears once per screen

In Slack’s dense aubergine UI, one accent is enough. Mint marks only the Day One row, so it’s findable without competing for attention.

08REFLECTION

The best decision we made was letting the research change our mind.

We could have shipped the first idea. Instead we followed the interviews to a sharper problem, and the product got quieter, more private, more useful.

What we learned

  • The barrier to belonging is information, not personality. People go quiet because they don’t know the rules, not because they’re shy.
  • Framing matters twice: “anxiety” belongs in the research, but never on the screen the anxious person is reading.
  • A privacy-first data model isn’t a constraint on the design. It became the reason people would trust the product at all.

Weeks of silent observation, shrunk to a first day.

SOURCES

Onboarding: Gallup, State of the Global Workplace, cited via StrongDM “Employee Onboarding Statistics.” Peer support: randomized peer-support intervention, PMC/NIH (PMC9358944). Repeated exposure: meta-analysis of 208 experiments on exposure and confidence in unfamiliar settings.