At Unovative, we believe technology should be inclusive, accessible, and meaningful for everyone.
Most mobile applications are designed around an invisible assumption: the user already knows how to use a smartphone.
They know what a bottom navigation bar does. They understand icons. They know how to recover after tapping the wrong button. They are comfortable typing, navigating menus, and changing settings.
For many senior citizens, those assumptions become barriers.
Families across India are increasingly giving smartphones to their parents and grandparents with the hope that technology will help them stay connected, organised, and independent. But having a smartphone does not necessarily mean feeling comfortable using it.
At Unovative, we didn't want to build another simplified version of an existing application.
We wanted to build something that adapts to the person using it.
We called it Sahayak - Hindi for helper.
The Problem We Set Out to Solve
Once we stopped assuming that everyone uses a phone in the same way, the friction became obvious.
Senior citizens can encounter challenges such as:
- Small buttons and dense screens that are difficult to interact with.
- Text that becomes harder to read as eyesight changes.
- Multi-step workflows that create unnecessary cognitive load.
- Typing that feels unfamiliar or uncomfortable.
- Fear of deleting something, pressing the wrong control, or encountering an error that they don't know how to recover from.
The problem wasn't simply:
"How do we build an AI chatbot?"
The real question was:
How do we make technology feel simple, approachable, and useful for someone who hasn't spent a decade building muscle memory around smartphones?
That question became the foundation for Sahayak.
What Is Sahayak?
Sahayak is Unovative's AI-powered mobile companion for senior citizens.
It runs on Android and iOS, with voice as its primary interface.
Instead of requiring users to navigate through complicated screens, Sahayak allows them to communicate naturally:
Speak → Sahayak understands → Sahayak responds
The response is available both as spoken audio and clear on-screen text, giving users two ways to consume the information.
Built for India
Sahayak supports:
- 🇮🇳 English
- 🇮🇳 Hindi
- 🇮🇳 Marathi
Language isn't treated as an afterthought. Speech recognition and spoken responses follow the language the user is most comfortable with.
What Can Sahayak Help With?
Sahayak is designed around everyday needs rather than complicated application workflows.
Everyday Questions
Users can ask questions naturally and receive short, easy-to-understand answers.
Smart Reminders
Users can create reminders for:
- 💊 Medicines
- 🏥 Appointments
- 🍽️ Meals
- 💧 Hydration
- 📋 Daily tasks
Instead of navigating through multiple settings screens, the user can simply speak.
"Remind me to take my medicine at 8 PM."
Conversation History
Previous conversations remain accessible so useful information can be revisited without asking the same question again.
Quick Actions
Common actions are available directly from the home screen so users don't have to search through multiple menus.
Easy Access to Help
When assistance is needed, the application provides a straightforward way to reach the relevant help.
Why Voice Comes First
Voice is the interface most people already know how to use.
Someone who may never type:
"How do I make tea?"
can simply say it.
Someone who may never find a reminder settings screen can say:
"Remind me to take my medicine at 8 PM."
One spoken instruction can replace an entire chain of taps.
Voice also changes what a mistake feels like.
A wrong tap can take a user to an unfamiliar screen.
A misunderstood voice request can simply be repeated or clarified.
That makes the cost of making a mistake much lower.
And for Sahayak, reducing the fear of making mistakes is an important part of building confidence.
Designing for the Person, Not the Pattern
Designing for senior citizens changed almost every decision we made.
1. Larger Typography
Text needs to remain readable without forcing users to zoom or struggle with small UI elements.
2. Clear Visual Hierarchy
Instead of presenting several equally important actions, screens focus on a clear primary action.
3. Generous Touch Targets
Buttons and interactive elements are designed to be easier to identify and tap.
4. Predictable Navigation
The user should always understand where they are and how to go back.
5. Plain Language
Error messages and instructions should explain what happened in language a person can understand.
6. Reduced Cognitive Load
We treated cognitive load almost like a budget.
If a three-step interaction can become a single spoken instruction, we should remove those three steps.
Accessibility is not a feature we add at the end. It is a constraint that shapes the product.
A larger microphone button isn't simply an accessibility accommodation.
It's good design.
How a Conversation Works
The interaction is intentionally linear.
The user shouldn't need to know which screen they are currently on.
The basic flow
User
↓
Open Sahayak
↓
Tap Microphone
↓
Speak
↓
Speech Recognition
↓
AI / Intent Understanding
↓
Generate Response
↓
Text Response + Voice ResponseStep 1 Listening
The application records the request and provides visual feedback that it is listening.
Step 2 Speech Recognition
The user's speech is converted into text using models designed to handle Indian languages, including English, Hindi, Marathi, and Indian-accented English.
Step 3 Understanding the Request
The system determines what the user is trying to accomplish.
The request could be:
- A question
- A reminder
- A navigation request
- A simple conversation
Step 4 Generate a Response
The response is intentionally designed for voice.
That means:
- Short sentences
- Simple language
- Minimal jargon
- No unnecessary walls of text
Step 5 Speak and Display
The response is spoken back to the user and displayed on screen.
This creates an important fallback:
If the audio is difficult to hear, the user can still read the response.
The Technology Behind Sahayak
We chose a technology stack that allowed us to support Android and iOS, iterate quickly, and provide reliable Indian-language voice interaction.
The goal wasn't to use the most technologies possible.
The goal was to choose technologies that supported the user experience we wanted to create.
Architecture
At a high level, Sahayak connects the mobile experience with AI, storage, and supporting services.
flowchart TD
U[Senior Citizen] --> APP[Sahayak Mobile App]
APP --> VOICE[Voice Input]
VOICE --> STT[Speech Recognition]
STT --> AI[AI / Intent Processing]
AI --> DB[(Supabase / PostgreSQL)]
AI --> RESPONSE[Response Generation]
RESPONSE --> TTS[Text-to-Speech]
RESPONSE --> APP
TTS --> APP
APP --> STORAGE[AWS S3]The mobile application and APIs live together so that the voice pipeline can be developed as one system.
Sahayak is primarily designed around short sessions, one clear task, and a response the user can trust.
How We Approached the Build
01 Understand the Day Before Designing the Interface
Before thinking about screens, we focused on what an elderly user actually needs and how they would naturally ask for it.
That led us toward:
- Voice-first interaction
- Short sessions
- Simple workflows
- Reliability over novelty
02 Establish Senior-First Design Rules
Certain decisions were made early:
- Typography
- Colour
- Spacing
- Large touch targets
- Simple navigation
- Minimal onboarding
The onboarding process asks only for information that matters, such as:
- Language
- Name
- Age range
- Gender
Age is not treated as purely demographic information. It can influence how quickly voice responses are delivered so that older users can hear a calmer, more comfortable response.
03 Make Listening Visible
When someone speaks to an application, one of the biggest uncertainties is:
"Is the application actually listening to me?"
Sahayak uses a live waveform while recording.
It is a small interaction detail, but it provides immediate feedback and removes uncertainty.
04 Assume Every AI Step Can Fail
AI systems aren't a single operation.
Voice recognition can fail.
The AI model can be slow.
Text-to-speech can fail.
Network connectivity can disappear.
So we designed the system with the assumption that every stage can fail independently.
For example:
Speech Input
↓
Speech Recognition
↓
AI Processing
↓
Text Response
↓
Text-to-SpeechIf text-to-speech fails, the text response is already available.
The interaction doesn't need to become:
"Something went wrong."
Instead:
The user still gets the answer.
Keeping Conversations Together
Conversation history is an important part of the experience.
Related questions remain accessible in History so that users can find useful information again without having to repeat the entire interaction.
This is especially important for an application designed around assistance.
A useful answer shouldn't disappear just because the conversation ended.
Where Sahayak Is Meant to Help
🏥 Healthcare & Elder Care
Users can ask questions about routines or symptoms and receive short, careful responses.
For health-related decisions, Sahayak should complement not replace professional medical advice.
💊 Daily Reminders
Users can create reminders for:
- Medicines
- Appointments
- Meals
- Hydration
- Daily tasks
All through natural interaction rather than complicated settings.
👨👩👧 Family Caregiving
Conversation history can help families understand routines and provide support without constantly taking control of the device.
🌐 Digital Inclusion
Sahayak can provide a simpler way for senior residents and first-time smartphone users to access information without first learning how every part of a modern smartphone works.
The Hardest Problems We Faced
1. Confidence, Not Just Usability
Traditional usability asks:
"Can the user complete the task?"
For our audience, we found a more important question:
"Does the user feel safe attempting the task?"
That changes design decisions.
Destructive actions need deliberate interaction.
Errors need to recover cleanly.
And when voice audio fails, the written answer should remain available.
2. Voice on Real Indian Networks
A spoken interaction can touch several different services.
Voice Input
↓
Speech Recognition
↓
AI Processing
↓
Response Generation
↓
Text-to-Speech
↓
Audio ResponseEvery additional step creates another opportunity for latency or failure.
That's why fallbacks aren't an engineering afterthought.
They're part of the product.
Our principle became:
The written response is the floor. Spoken audio is the enhancement.
3. Indian Languages in the Real World
Generic voice engines aren't always designed for the realities of Indian speech.
Hindi, Marathi, English, and mixed-language conversations introduce additional complexity.
Choosing technology designed specifically for Indian languages therefore became a product decision, not simply a technical one.
4. Authentication Without Friction
Traditional authentication flows often require:
- Phone numbers
- OTPs
- Passwords
- Multiple verification steps
For our target audience, every additional screen creates another opportunity to lose the user.
Device-based authentication removes much of that friction.
For this audience, the login screen can be where the product dies.
What We Learned
Simplicity Is Harder Than Adding Features
Adding another feature is often easier than removing a step.
When we remove complexity from the interface, that complexity doesn't disappear.
It has to move somewhere else usually into the architecture.
The user shouldn't have to see it.
AI That Sometimes Fails Is a Feature That Sometimes Fails
AI systems are probabilistic.
Voice systems fail.
Networks fail.
Services become unavailable.
The solution isn't pretending that these things won't happen.
The solution is designing the product so that one failed component doesn't destroy the entire interaction.
Voice Changes UX
Voice doesn't have a traditional menu to browse.
The system has to do the work that buttons normally do:
- Understand the user's intent.
- Determine what they need.
- Respond concisely.
- Speak naturally.
- Recover when the request isn't clear.
This makes conversational design an important part of the product.
Designing for a Constrained User Makes Products Better for Everyone
Large touch targets.
Short answers.
Clear navigation.
Calm error messages.
Simple workflows.
These aren't improvements that only benefit senior citizens.
They make applications easier for everyone to use.
What's Next for Sahayak?
Sahayak is actively evolving.
The foundation today is:
- Voice assistance
- Smart reminders
- Conversation history
- Senior-first interface
The next areas we're exploring include:
Better Reminder Delivery
Improving how reminders are delivered and managed.
Persistent Preferences
Making preferences such as language and voice speed persist consistently throughout the application.
Deeper Accessibility
Applying accessibility choices across the entire experience rather than treating them as isolated features.
Caregiver Experience
Expanding the caregiver experience so families can support senior citizens without taking control away from them.
Mixed Hindi-English Speech
Supporting the way people naturally communicate switching between Hindi and English within the same conversation.
Better Weak-Network Support
Making essential assistance more resilient when network connectivity is unreliable.
The Bigger Idea Behind Sahayak
Sahayak isn't simply a chatbot wrapped in a mobile application.
It is an attempt to answer a simple but difficult question:
What does it take for technology to be genuinely useful to someone who has spent most of their life without it?
The answer is mostly about what we remove.
Remove the multi-step workflow.
Remove the typing requirement.
Remove the assumption that the person already knows how the phone works.
Remove the error message that explains nothing.
Remove the fear of making a mistake.
What remains should be simple:
Listen. Understand. Reply. Remember.
Closing
At Unovative, we measure technology not only by what it can do, but also by who can use it.
Building Sahayak brought together mobile engineering, AI, voice technology, accessibility, and human-centred design around one goal:
Build technology that works for the person using it.
For us, Sahayak represents a broader belief:
The future of AI shouldn't only be intelligent. It should be inclusive.
If you're working on healthcare, elder care, accessibility, or digital inclusion, we'd love to share what we're learning as we continue building Sahayak.