Independent Product Design Exploration

Goodwill

Civic disruption is tolerable. Civic surprise is not.

I noticed something simple, repeatedly: civic work in Bangalore, pipe repairs, road digging, power changes, happens to people, not with them. I couldn't talk to thousands of residents or sit inside city departments, so I used HMHT, my framework for designing when real human access isn't possible, end to end.

Role
Sole product thinker and designer, self-initiated
Timeline
4 weeks, full HMHT cycle
Domain
Civic infrastructure transparency, Bangalore
Status
In preparation for a formal proposal at Tarento
01
The problem

The city stopped explaining itself, and residents stopped trusting it.

Roads close overnight without warning. Pipes get dug up at 7 AM on a workday. A repair that should take three days runs into its third week with no explanation. Residents, including me, find out by encountering the work, not before it.

The information isn't missing, it sits in tender documents, press releases, websites nobody visits. It just doesn't reach the people whose lives it disrupts, in a form they can use, when it would help. That gap is a slow, daily erosion of trust, and it hardens into a working theory: nobody is in charge, or nobody cares.

“Frustrated, then resigned. The frustration is short-lived. The resignation accumulates, until resignation is the default civic emotion.” A composite voice, synthesised from HMHT simulation, not a real quote.

Source: lived experience as a Bangalore resident, informal observation across HSR Layout, Indiranagar, and Jayanagar, and publicly visible civic activity through 2025–2026, cross-checked against TRAI, NPCI, and BBMP/BMRCL public reporting.

02
Approach

Two decisions that carried the whole case.

HMHT ran in full, five phases, Human Simulation, Behavioural Mapping, Information Architecture, Interaction Direction, and UI Generation, each producing an artifact that fed the next. Phase one alone produced four personas, a commuter, a shopkeeper, a civic-minded resident, and the field officer whose two-tap update has to work in direct sun, each clearly marked HMHT-simulated, not real research.

Decision 01 · Load-bearing
Reframed the brief from a notification system to a trust-restoration system.
The obvious framing, “how do we communicate roadwork better,” produces another notification system, judged by how many messages it sends. The actual problem was “how do we restore civic trust through information design,” which produces a system judged by predictability, whether residents stop abandoning trips they didn't need to. Notifications became supporting infrastructure, not the product itself.
Decision 02
Set four commitments, each one a thing explicitly rejected.
Clarity over completeness, three sentences, not three paragraphs, rejecting a tender document dumped into a UI. Transparency over polish, “we don't know yet” beats a confident estimate that turns out wrong, rejecting false certainty. Local before global, a street's residents hear first, rejecting broad announcements styled as targeted ones. Officer-first input, an update takes under thirty seconds in the field, rejecting a resident-only design that assumes the input pipeline holds on its own.
Decision 03
Built a three-node architecture with a named latency target.
Field Officer, an Operational Truth Layer, and Resident, one operational truth, translated automatically into six channels: web, app push, WhatsApp, SMS, IVR, and a printable ward-board with a QR code back to the source page. The target: an officer's update at 11:14 becomes a resident's notification within the same minute.
Goodwill, Resident Home, Project Page, and Officer Update screens

The three core screens. Status by colour, consistently, vermillion for active, amber for delayed, forest for on-schedule. System-written text and human-written text are typographically distinct throughout, so the voice is never ambiguous.

03
Outcome

Where the work actually stands.

4 weeks
Full HMHT cycle, human simulation through UI generation, each phase's output feeding the next.
6 channels
Web, app, WhatsApp, SMS, IVR, and a printable ward-board, all rendered from one operational truth layer.
<60 seconds
Target latency from an officer's field update to a resident's notification, by design, not yet measured live.
Not yet validated
Concept testing, usability sessions, and adoption tracking are scoped but haven't run, staged honestly as next, not claimed as done.
04
Reflection

I thought I was designing a better notification system. Midway through simulating a resident in her late fifties, the simulation surfaced a moment I hadn't anticipated: she missed her granddaughter's school event because an auto-rickshaw couldn't get through, and no one had told her that morning. That's not a problem a notification system solves. It's solved by treating someone's time and dignity as worth the effort of being told.

What's here is the case, end to end. A more detailed walkthrough, personas, journey maps, service blueprint, stakeholder map, is available in conversation.

Request a walkthrough