All 5 had a personal workaround. All 5 named Slack primary. A capacity check took 6.5 min and 3+ tool switches. 2 gave up. 4 of 5 would cancel a recurring meeting. Changed: Dunio has to be where work happens, not a window onto it.
Dunio · Freelance · 2026
Radhika rebuilds a Notion page every morning because Jira can’t tell her what’s going on. I designed the thing that should delete that page.

Role
Product designer, end to end
Timeline
3 weeks research, 7 weeks design
Client
A 20-person operations and delivery consultancy, Bangalore
Status
Designed and handed to engineering, in build
Problem
8 to 12tools per teamSurvey, n=~80
~1.5hlost per dayObserved, 5 teams
68%name tool fatigue as pain #1
70%have no single source of truth
Information exists everywhere. The context connecting it is fragmented across projects, conversations, documents, calendars and people.
Not one of those tools is broken. The failure is in the space between them, and nobody owns that space.
The solution
Dunio is one layer over the tools a team already uses: projects, people, schedules and chat in one place, loaded by role.
Ten menus, one dashboard, and chat built in, so the work and the conversation about it stay together.
Before. Nine tabs, six systems, one morning.After. One dashboard, loaded by role.
Client asked
“Build a better PM tool for enterprise teams.”
I pushed it to
“Build the layer that connects what teams already use.”
The first brief puts us against Jira, where a 70-person firm loses. The second puts us where nobody is selling.
Field notes
0interviews
~80surveys across 8 companies
0hours shadowing
0competitor tools, run for a week each
Six things people kept saying
Hover or tap a card to hear it in their words.
Three people I designed for
Composites from the 8 interviews. Each has a success signal, the thing that has to stop happening for the design to have worked.
“That’s not management, that’s manual data entry with an audience.”
Wins whenthe weekly status sync gets cancelled
Radhika’s Tuesday morning
Why the existing tools didn’t close it
AsanaPeople kept the plan here and the conversation in Slack.
Plot twist
Week 2 me wrote “we do not compete with Slack” in the competitive summary.
Week 4 me spent six weeks of engineering budget building chat into the product.
The two-week test that changed my mind is in The calls →
The bet
How might weHow might we give a team full operational clarity across projects, people, schedules and communication?
without context-switchingwithout manual assemblywithout one more system to adopt
Rule 01If a screen adds information without removing a question, it doesn’t ship.
Out of scope on purpose
Replacing JiraDocument editingPayroll and HRExternal client comms
Each is a real need and each makes the product worse at its only job.PhasesOne screen replaces four status meetings.
Radhika assembles the week from six systems.
15 to 10
Card sort, 8 participants. “Where’s my team’s schedule?” got seven different answers from eight people. That’s structural, not a labelling problem, so I started again.
15modules in the nav
ProjectsTasksGoalsDocumentsPeopleDepartmentsChatCalendarAnalyticsAdminSettingsIntegrationsReportsNotificationsTime Tracking
Moved inside the work
Five modules didn’t get deleted. They moved inside the work, which is the whole thesis expressed as a nav bar.
Three directions
A
Module-first
Rejected. It rebuilds the same fragmentation inside one product.
B
Fully customisable canvas
Tested and rejected. 3 of 5 abandoned setup after 9 minutes. “This is like setting up a new phone.”
C
Project-centric, role-based defaults
Chosen. The project is the container. Your role decides what loads.
What broke
01Round 2I assumed chat was someone else’s job.3 of 5 went hunting for a Chat section. I had designed a product people were supposed to leave in order to talk about it.
02Round 2I gave people a blank canvas and called it flexibility.3 of 5 abandoned setup after 9 minutes. Configurability is a cost you charge the user on day one, before they trust you.
03Round 3I designed a Scheduler that doesn’t know what Projects knows.I scoped it to v1.1 instead of fixing it, which was the wrong call.
01
“I didn’t open Slack once during that session. I don’t think that’s ever happened in a work context.”
P
Participant
Round 3
02
“This is exactly the dashboard I build manually every Tuesday morning.”
P
Participant
Round 3
Five people per round. Directional, not proof. These are signals I acted on, not results I’m claiming.
The calls
Hover or tap a call to see the receipt.
Where we disagreed
What I cut
Predictive blocker detection. Needs 90+ days of real usage data to say anything that isn’t a guess.
Cross-company benchmarking. The contracts alone outran the project. Four-month opt-in build.
Collaborative task editing. Depended on the architecture finance killed. This one still bothers me.
The product
Eight screens carry the whole argument. Each one exists because of a decision above it.
Ten menus. Pick one to see what’s inside.
01 of 10The morning view: how the team and the projects are doing.InsideTeam productivityPerformance by week, month, yearProject completion statusTask distribution
01
One screen in place of four status meetings.
Nothing here is configurable, and that is the feature. Your role decided what loaded.
02
The decision, the file and the conversation live on the work.
This is “five modules moved inside the work” made visible.
04
50+ permission toggles became three roles.
All 5 understood it in 3 minutes, and admin setup dropped from 8-hour cycles to under 22 minutes.
05
Every department, its people and its projects, in one list.
Director, headcount, project count and status at a glance.
06
Chat, calls and files without leaving Dunio.
Direct messages, channels, shared files, and audio or video calls from the same thread.
07
Analytics, designed to answer a question.
It ships as standard reports for now. The why is in The gap.
08
Settings, kept out of the way.
Seven tabs: Profile, Notifications, Security, Appearance, Integrations, API and Advanced. Three of them do the heavy lifting.
Two-factor sign-in with an authenticator app, password and login rules, and a list of active sessions you can revoke.
What it serveskeeping the account safe
Walkthrough
Everything else, in one recording.
Every screen and flow, start to finish, including the ones that didn’t fit on this page. All edge cases covered, with all the remaining features.
Creating a project and assigning the team
1Enter Projects
2Create Project
3Add Details
4Add Team
5Set Permissions
6Share Link
7Team Notified
The two critical flows I stress-tested.
Design system
Duniofied. Built for this project.
Poppins, a token set, and the components behind every screen above. Purple is reserved and unused, held for the v2 automation layer.
The gap
I handed this over on a Friday. This is the honest column-by-column of what survived.
DesignedAction-oriented, every chart answers a question
Going to buildStandard reports
WhyThe timeline ran out. The one I’m least proud of. It ships as the thing I spent the project arguing against.
The Scheduler that goes into the build
Outcomes
2.3 minto create a task with full context
7.5 minto first meaningful action, unassisted
3 of 5closed Slack unprompted
<22 minadmin setup, against 8-hour cycles observed in research
And the one that actually matters: does Radhika delete her Notion doc?Nobody can answer that until this is in front of a real team on a real Tuesday.
Next
Scheduler ↔ Projects two-way sync, v1.1
AI status inference, roughly 60% pattern-match today, not good enough to ship
Google Calendar and Outlook sync
Team pulse surveys
What I’d change
01
I tested individuals, and this product only exists between people.
Every session was one person alone with a screen. The premise is coordination, the one thing a solo session cannot show you. Next time: three people from one team, same session.
02
I shipped Analytics knowing it was wrong.
I knew in week 5. I raised it in week 7. Two weeks earlier and it would have been a scope conversation instead of a casualty.
03
The Scheduler gap should have delayed the launch by two weeks.
I had the evidence in round 3 and let it go to build because the handover date felt fixed. Dates are rarely as fixed as they feel on the Thursday before.
04
Engineering shaped this design more than any user did.
One line in a cost model removed a feature, changed an architecture and killed collaborative editing. Constraints don’t arrive at the end of design. They arrive in the middle.














