Overview - evolution of chatbot
From Deterministic Chatbot to AI-Powered Assistant
I transformed Verizon’s chatbot from a deterministic, UI-led experience into an AI-powered dynamic assistant. The shift moved the experience beyond predefined flows and rigid UI patterns toward conversations that could understand user intent, adapt to context, and respond more naturally - making the assistant smarter, more intuitive, and more capable of supporting end-to-end service journeys.
Conversational AI
Product Design
UX Strategy
Timeline
Jan - May 2026
Evolution across multiple phases
MY ROLE
Defined the experience strategy, interaction model, and conversational patterns.
team
Product · Design · Legal · VDS · Content · Business · Engineering
IMPACT
From scripted interactions to dynamic, AI-powered service journeys across upgrades, plan changes, and billing.
phases
GEN 1 · STARTING POINT
The Reactive Chatbot
A basic chatbot that provided links to Verizon.com pages or FAQ articles. Keyword-based routing with static prompts meant users had to navigate themselves to find answers.
GEN 2 · UNIFIED CHAT
The Journey Engine
Introduced full customer journeys directly inside the chat: Plan Change, Device Upgrade, Add a New Line, Home Internet. Handled all edge cases, comprehensive error handling, and seamless agent handover when needed. Reusable journey templates enabled rapid scaling.
GEN 3 · INTELLIGENT FUTURE
The Intelligent Ecosystem
AI that is fully context-driven. Anticipates customer needs, handles all types of inquiries — not just pre-built journeys. Understands intent, resolves issues proactively, and escalates gracefully to live agents when needed. Every conversation reaches resolution.
Impact
From Smarter Conversations to Measurable Outcomes
From a rule-based engine (2024) to an AI-powered conversational platform (2026), the assistant evolved from scripted interactions to dynamic, context-aware experiences.
challenge
What this project set out to solve.
How might we help Verizon customers navigate complex, multi-step sales journeys by providing instant, guided support at every step, so they never feel lost?
research
Competitor Scan
Comparing leading conversational assistants to identify patterns, gaps, and opportunities for Verizon.
Emerging Trend References
Emerging patterns influencing how people interact with AI assistants.
RCS & Call Centre Agent Research
Research with call-centre agents and RCS models helped identify key opportunities for improving the Verizon Assistant experience.
Heuristic Evaluation
Evaluated key user journeys to identify usability issues, friction points, and opportunities for improvement.
Hover over to magnify details
Cross-functional Workshop
Collaborated with product, Home-sales website, design, engineering, VDS, Content and AI/ML teams to align on challenges, opportunities, and the future experience.
Exploration
Defining the AI Assistant Experience
We explored four key dimensions of the experience—from how intelligence is expressed to how users enter, navigate, and interact with the assistant.
Theme 1: Signaling Intelligence (Motion)
How should the assistant communicate that it's thinking? The four concepts testing how the interface communicates "I'm processing this.
- Sparkle icon + red accent - a dedicated animated icon above the input
- Directional gradient motion- the background itself shifts to indicate activity
- Rotating border - a processing indicator around the container edge
- Text animation - the response text itself animates in
Theme 2: The Home Screen Problem (Layout)
How much content should the assistant surface before a conversation begins? Quick iterations testing how much content a "chat home" should carry before it stops feeling like chat.
- Empty state (just the input field)
- Single product/deal tile
- Dual-tile layout (trouble + order tracking)
- Full carousel with perks/deals
Theme 3: Getting Into a Conversation (Entry Points)
How should users choose how they want to get help?
- Combining search + Ask AI together.
- A digital concierge let's the users select how they want to go ahead inside chatbot.
Theme 4: Window Behavior (Container States)
How should the assistant behave as a companion?
- Companion mode variations — draggable/resizable vs. fixed 50%-height.
- Freely resizable + draggable
- Locked to 50% height only
Concepts
Exploring the Future Assistant Experience
Exploring different ways to make the assistant more accessible, contextual, and seamlessly integrated across Verizon experiences.
Future Concepts - Verizon App
I did some concepting around how chat could evolve, this fundamentally changes the entrypoint.
MY TAKEAWAY
Designing the assistant meant designing beyond the chat.
The biggest shift wasn't introducing AI—it was rethinking what an assistant could do when conversation, context, and action came together.
