Plain language water information
Respond to supported questions about drought, conservation, and general Arizona water topics with a concise explanation and source direction.
Back to Work
Public utility chatbot
AZ Water Chatbot: Ask Blue About Water in Arizona
A responsive conversational prototype that helps people ask plain language questions about Arizona water, drought, conservation, and the Colorado River, with visible emergency and language pathways.
AudiencePublic / Civic
DomainCivic design · Conversational UX
RoleProduct designer
PlatformResponsive web
Year2023
ScopeCase study and prototype
Reading time4 min
View the product
The prototype explores suggested questions, open text, emergency guidance, language switching, responsive behavior, and clear fallback states.
30 Second Summary
DESIGN RESPONSE
A conversational prototype with suggested questions, open text, direct emergency guidance, and a visible language switch.
EVIDENCE
Five participants, ages 20 to 35, tested four prototype tasks.
RESULT
The study clarified how direct actions, suggested prompts, and clear fallback language could reduce the findability problems observed in testing.
Overview
A plain language guide to Arizona water information.
The Arizona Water Chatbot is a responsive conversational prototype for drought, conservation, the Colorado River, emergency guidance, and language access. I designed the information pathway, interaction model, and tested prototype.
Problem
Technical information offered no obvious starting point.
Residents had to navigate institutional sources and unfamiliar terminology before finding a useful answer. The experience needed clear expectations, visible safety boundaries, direct actions, and an honest path to official sources.
Responsible boundary
The chatbot should answer, defer, or route based on the stakes.
The prototype supports information guidance, official source direction, language access, and a persistent emergency path. It does not show a human handoff screen.
Current policy or account specific detail
When the answer depends on a current official rule, utility account, or unavailable source, explain the limit rather than generating certainty.
Emergency or unsafe situations
Keep Emergency Help persistent and direct the user to the appropriate official or emergency source without attempting diagnosis.
Product evidence
Three states show how the conversation establishes trust.
The sequence sets expectations, shows an active exchange, and keeps the answer boundary inspectable. Language access remains a prototype affordance rather than proof of a complete translated experience.
Desktop question and answer
The complete desktop viewport keeps navigation, the active conversation, and the response boundary visible together.
Desktop emergency guidance
The emergency flow keeps the assistant limits clear and points people toward authoritative support.
Responsive proof
The desktop reference and mobile viewport remain visible together without a nested phone or captured recording field.
Usability testing and iteration
What five participants could find, and what the notes cannot prove
Five participants, ages 20 to 35, completed four tasks. The results below distinguish observed navigation from tasks whose wording or scoring was not fully recorded.
Emergency route: 5 of 5
All five participants found the persistent emergency route. The notes do not establish whether assistance was allowed.
Language switch: 5 of 5
All five participants found the language action. Their language background and Spanish comprehension were not documented, so this result covers findability only.
Study limits
The remaining task notes do not describe the activities precisely enough to support broader information retrieval claims.
Reflection
Fallbacks define the conversation
Conversational design taught me that failure states are half the product. A trustworthy chatbot is defined by how clearly it handles uncertainty, not only by how smoothly it answers supported questions. Designing the fallback, source boundary, and emergency path gave the success states their credibility.
Next project
Lilly Health
Lilly Health shows shipped healthcare product work across onboarding and reusable visualization systems.
