Harshika

Hire Me!

Harshika

Hire Me!

Harshika

Back

Back

Conversational AI chatbot experience for small business owners
Conversational AI chatbot experience for small business owners

Overview

Overview

Timeline
Aug 2024 - Dec 2024 (4 Months)


Target Audience
Small business owners who are first time users of the Salesforce website

Timeline
Aug 2024 - Dec 2024 (4 Months)


Target Audience
Small business owners who are first time users of the Salesforce website

Team
8 designers + researchers and 2 Salesforce mentors

My Contributions
Led the chatbot personality sprint (defined and conducted user research), storytelling and presentation. Contributed in brainstorming, ideating and designing 8 end-to-end chatbot features, and facilitated team & stakeholder meetings

Team
8 designers + researchers and 2 Salesforce mentors

My Contributions
Led the chatbot personality sprint (defined and conducted user research), storytelling and presentation. Contributed in brainstorming, ideating and designing 8 end-to-end chatbot features, and facilitated team & stakeholder meetings

TL;DR

Fin is an AI-powered chatbot my team and I designed for Salesforce to make product discovery feel effortless. Instead of pushy sales tactics, Fin offers context-aware guidance, empathetic support, and clear comparisons, adapting to user behavior while still meeting business goals.

Key moments included shaping Fin’s personality, designing adaptive triggers, simplifying product comparisons, and creating respectful ways to re-engage when users go quiet.

TL;DR

Fin is an AI-powered chatbot my team and I designed for Salesforce to make product discovery feel effortless. Instead of pushy sales tactics, Fin offers context-aware guidance, empathetic support, and clear comparisons, adapting to user behavior while still meeting business goals.

Key moments included shaping Fin’s personality, designing adaptive triggers, simplifying product comparisons, and creating respectful ways to re-engage when users go quiet.

Over four months, I co-led a team of eight designers and researchers to define and design an improved chatbot experience

Over four months, I co-led a team of eight designers and researchers to define and design an improved chatbot experience

Over four months, I co-led a team of eight designers and researchers to define and design an improved chatbot experience

35+

35+

Collaborative sessions

Collaborative sessions

20+

20+

Critiques received from UX team

Critiques received from UX team

8+

8+

Use cases designed

Use cases designed

ONE OVERARCHING QUESTION

ONE OVERARCHING QUESTION

How might we design a human-centered AI chatbot experience for small business owners
How might we design a human-centered AI chatbot experience for small business owners
How might we design a human-centered AI chatbot experience for small business owners

Problem Statement

Pushing to Contact Sales

Pushing to Contact Sales

Poor Product Discovery

Poor Product Discovery

Being

Too Robotic

Being Too Robotic

Endless Scrolling

Key Issues

Key Issues

Solution
a.k.a the juicy part

A number of use-cases have been solved for based on our primary research. From an overarching personality of the chatbot to triggering chatbot responses on certain user behavior.

Personality

Goals for the personality design

Improving user engagement with the chatbot through language and tone

Increasing user retention

Designing a seamless bot to human(Sales-rep) transition

Comparing Products

Pain Point


Challenging for users to compare and retrieve product details in the chatbot conversation


Key Insight


Users expect the chatbot to simplify product discovery by providing quick information to help them decide whether to explore further


Solution


Summarize product key features and create comparison table within the chat

Chat Timeline Progress Bar

Pain Point


Endless scrolling to find specific information in the chat


Key Insight


Users want an easier way to reference relevant info such as product suggestions in chat without scrolling endlessly


Solution


Real-time timelines of the chat


Flexibly open or close the timeline

Progress indicators while scrolling the chat


Titles on timeline are created in real-time

Chatbot Redirection

Pain Points


Users worry they will lose the chat upon switching web pages


Key Insight


Users need assurance that their chat and shared content (like product links) are safe and accessible while browsing


Solution


Chatbot Redirection; Helps the users do a dynamic search of different products on the website, by leveraging chatbot tools

Behavioral Triggers

Users often input one word entries and expect relevant responses by an AI Chatbot

How might we: design the chatbot to prompt users for clarification or elaboration?
and ensure they feel confident it will understand detailed responses?

Clarifying Questions

Suggestive Prompts

Connect to Agent

User Inactivity

Behavioral Triggers

Users often input one word entries and expect relevant responses by an AI Chatbot

How might we: design the chatbot to prompt users for clarification or elaboration?
and ensure they feel confident it will understand detailed responses?

Clarifying Questions

Suggestive Prompts

Connect to Agent

User Inactivity

Behavioral Triggers

Users often input one word entries and expect relevant responses by an AI Chatbot

How might we: design the chatbot to prompt users for clarification or elaboration?
and ensure they feel confident it will understand detailed responses?

Clarifying Questions

Suggestive Prompts

Connect to Agent

User Inactivity

Behavioral Triggers

Users often input one word entries and expect relevant responses by an AI Chatbot

How might we: design the chatbot to prompt users for clarification or elaboration?
and ensure they feel confident it will understand detailed responses?

Clarifying Questions

Suggestive Prompts

Connect to Agent

User Inactivity

15+

Contextual Inquiries

156.8K

Projected small-business sessions improved

10+

Design

Iterations

Now onto the messy part

that got us the juice

Now onto the messy part

that got us the juice

A PECULIAR CHALLENGE

A PECULIAR CHALLENGE

We needed a way to observe users' product-procurement chatbot interaction and test personality preferences, even though no bot existed that supported both.

We needed a way to observe users' product-procurement chatbot interaction and test personality preferences, even though no bot existed that supported both.

I suggested leveraging GPT to simulate Einstein AI for conducting contextual inquiries

I suggested leveraging GPT to simulate Einstein AI for conducting contextual inquiries

This was to better understand users' preference of personality, tone, and assistance with product discovery before any human intervention

This was to better understand users' preference of personality, tone, and assistance with product discovery before any human intervention

Why GPT?

Why GPT?

The team was considering between Co-Pilot and GPT but the latter stood out for its consistency in responses in multiple conversations and devices.


It gave us the flexibility to personalize conversations and adaptability to test with participants from various industries.

The team was considering between Co-Pilot and GPT but the latter stood out for its consistency in responses in multiple conversations and devices.


It gave us the flexibility to personalize conversations and adaptability to test with participants from various industries.

Inspired by the 16 personalities framework we narrowed down to 3 personality types and had the AI model mimic them while responding to participants on their queries

Inspired by the 16 personalities framework we narrowed down to 3 personality types and had the AI model mimic them while responding to participants on their queries

How did we test?

How did we test?

Fun/Joy Personality

Fun/Joy Personality

Formal Personality

Formal Personality

Empathetic Personality

Empathetic Personality

Key Insights

Key Insights

Users value a detail oriented chatbot with actionable insights

Users value a detail oriented chatbot with actionable insights

They want a bot proactive in anticipating needs and be empathetic, inquisitive & clear

They want a bot proactive in anticipating needs and be empathetic, inquisitive & clear

No Jargons unless absolutely necessary with a formal tone

No Jargons unless absolutely necessary with a formal tone

3 Characters Based on Insights

Curious but Cautious

Curious but Cautious

Highly adaptable

Highly adaptable

High EQ, Sociable, & Helpful

The decision behind character creation

It was an opportunity to create not just a character but a joyful connection between the users and the chatbot. Characters are also a big part of the Salesforce branding and it'd only make sense to have a character for the chatbot to easily merge with the ecosystem

Introducing

Fin the Dolphin

Fin the Dolphin

Fin the Dolphin

Fin the Dolphin

In a parallel sprint, we studied product discovery and context awareness through contextual inquiries

Initial sketches & iterations

Feedback from Salesforce

Worked on 2 more rounds of iterations

based on continuous feedback from Salesforce, peers and internal brainstorming

based on continuous feedback from Salesforce, peers and internal brainstorming

User Testing

User Testing

User Testing

with mid-fidelity prototypes to test how much guidance vs. control users need in a conversational, business-oriented chatbot

with mid-fidelity prototypes to test how much guidance vs. control users need in a conversational, business-oriented chatbot

Final Design

Final Design

Final Design

after feedback from user testing, we got to the final solution

My takeaways from this project 🚀

Designing in larger teams is less about alignment and more about shared clarity

Designing in larger teams is less about alignment and more about shared clarity

Working in an 8+ person team taught me that velocity doesn’t come from everyone agreeing, but from everyone understanding the same problem framing. Clear artifacts, explicit assumptions, and well-scoped decisions mattered more than individual brilliance.

Working in an 8+ person team taught me that velocity doesn’t come from everyone agreeing, but from everyone understanding the same problem framing. Clear artifacts, explicit assumptions, and well-scoped decisions mattered more than individual brilliance.

01

Constraints are where meaningful innovation actually happens

Constraints are where meaningful innovation actually happens

Time pressure, limited access to users, and ambiguous requirements forced us to find creative workarounds, like leveraging AI to simulate or extend contextual inquiries. I learned that novel solutions often emerge not despite constraints, but because of them.

Time pressure, limited access to users, and ambiguous requirements forced us to find creative workarounds, like leveraging AI to simulate or extend contextual inquiries. I learned that novel solutions often emerge not despite constraints, but because of them.

03

Agile is only effective when it’s treated as a learning system, not a delivery system

Agile is only effective when it’s treated as a learning system, not a delivery system

Sprints were most valuable when we used them to test assumptions, close knowledge gaps, and make decisions, not just to ship outputs. This reframed agility for me as a way to reduce uncertainty continuously, rather than a process to follow rigidly.

Sprints were most valuable when we used them to test assumptions, close knowledge gaps, and make decisions, not just to ship outputs. This reframed agility for me as a way to reduce uncertainty continuously, rather than a process to follow rigidly.

02

Good problem-solving requires comfort with ambiguity before confidence in solutions

Good problem-solving requires comfort with ambiguity before confidence in solutions

Many of our hardest challenges didn’t have precedents or best practices. Sitting with uncertainty, testing imperfect ideas, and iterating in public with the team strengthened my ability to reason through novel problems instead of defaulting to familiar patterns.

Many of our hardest challenges didn’t have precedents or best practices. Sitting with uncertainty, testing imperfect ideas, and iterating in public with the team strengthened my ability to reason through novel problems instead of defaulting to familiar patterns.

04

and probably the most cliched but extremely valuable

Great team work truly makes the dream work💫

and probably the most cliched but extremely valuable

Great team work truly makes the dream work💫

Harshika Rawal

2025 💡

Made with Love❤️and Coffee☕

Harshika Rawal

2025 💡

Made with Love❤️and Coffee☕

Harshika Rawal

2025 💡

Made with Love❤️and Coffee☕

Harshika Rawal

2025 💡

Made with Love❤️and Coffee☕