From 0-1: Designing a unique chatbot builder platform

Simplifies customer service for small businesses by providing an intuitive, user-friendly solution.
B2B
End-to-end
SaaS Platform
2024
AI
Role
Product Designer
Team
Myself, CEO, product manager, multiple front-end and back-end engineers
Type
0-1, Branding, web app, product strategy, design thinking.
Duration
8 months
Tools
Figma

Overview

About ChatFusion

A ChatGPT-powered chatbot builder platform that enables businesses to easily create, train, and deploy custom chatbots using their own data. The platform aims to increase user engagement, reduce support costs, and provide personalized customer experiences at scale.

My Role

As the only designer on this project, I independently led the design of the entire product from 0-1 while working closely with stakeholders and product manager. My objective was to design a B2B product with B2C in mind, ensuring an easy-to-use product with data in mind.

Impact

With the MVP, the team has gotten several US companies to use the product while we continue to build the product and expand our users.

Business Challenge

Create an unique chatbot builder platform that provide unique values to users to make us stand out from the highly competitive chatbot market and to get users to sign up to our platform.

Understand

Our Users

Our product is targeted at small businesses struggling with limited human resources and in need of smarter chatbots to manage personalized customers interaction, ultimately growing their businesses.

Pain points

Through research and studies, we noted the different struggles between having a human agent vs. having an AI chatbot. Although having a AI chatbot could solve the problem for the costly human agent, the traditional AI chatbot still has its limits like limited understanding and training complexity.
We then tried to get deeper insights by talking to some of our existing clients and companies before identifying three main issues they faced while using AI chatbots. Based on these insights, we formulated strategies to solve the problems.
By leveraging ChatGPT model, we provide a chatbot capable of having a conversation and handling difficult tasks to improve customer satisfaction. By including an easy bot training process, we aim to solve the struggles for small businesses trying to create bots for their products.

Our Competitors

To stand out from the highly competitive and ever-changing market, we also did in-depth market analysis and identified a gap in the market.
By analyzing the current chatbot market, we identified a critical limitation in existing platforms: they operate on a single-bot architecture, requiring users to manually re-upload and re-train sources for every new bot they want to create. This repetitive process is inefficient, time-consuming, and creates unnecessary friction for businesses looking to scale their AI-powered communication solutions.

Product Strategy

Beyond driving leads and customers satisfaction, our platform recognizes the fundamental challenge facing small and medium businesses: the lack of technical expertise to develop and manage complex AI solutions. By creating a no-code, intuitive chatbot builder, we're enabling businesses with limited technical resources to leverage cutting-edge technology effortlessly and effectively.

Features

We organized the many features according to what might drive the biggest impact to our users:
1) Easy upload source for bot training
2) Ability to shared source across bots for easy implementation
3) Setting up lead generation and other objectives like signups
4) Manage bot performance like viewing analytics, leads and conversation history

Wireframe

As a new product, getting early stages feedback is crucial, especially for this ever-changing market needs. Thus, after planning our product features, we decided to launch a quick MVP based on our wireframes to test the product, to see how our features aligned with the user needs.
The testing focused on the main primary user journey of our platform:
Creating a bot and successfully train bot by uploading source

Design Solution

Highlight 1: Upload Your Own Source

By providing an easy upload source to train bot content, we aim to help businesses transform hours of manual content preparation and training into a matter of minutes. Enter a link or upload a file, and we will do the bot training for you.

Design Process: Learn more about designing the knowledge base and keeping experience consistent across the app here.

Highlight 2: Cross-Bot Sources Sharing

By setting knowledge bases to projects, we allows bots to share the same resources. With this, we've solved a critical pain point in chatbot development by implementing a project-centric knowledge base strategy. Our platform breaks away from the restrictive single-bot model, offering users the ability to create a centralized knowledge base that can be effortlessly shared across multiple chatbots.

Highlight 3: Friendly and intuitive bot setup

When doing competitor research, we noted the lack of live preview when setting up the bot, thus we made sure we include a side by side live preview while setting up the bot.

Design Process: Learn more about how I design the complex bot creation experience  here.

Highlight 4: Advanced Objective Options

Under the bot training tab, we provided businesses the option to add additional objective to the bot like collecting leads, prompting actions. This feature transforms the chatbot from a mere communication tool into a proactive business engagement platform, enabling companies to align their AI interactions with precise marketing and conversion goals.

Highlight 5: Friendly user experience for customers

Through multiple rounds of iterations, we decided on an option for business to choose between "Click to Chat" or "Auto-popup". The auto pop ups the welcome message and a quick messaging input field for customers to do a quick interaction before expanding into a full screen mode.

Thoughts

While the product is still under development, I felt like I have learned a lot from such a large scale project. I challenged myself to think about a lot of variations, created prototypes to test and to communicate within the team. Also to think about design from different users' mindsets, making thoughtful design decisions that benefit our users.

Design System

For design system, we used Ant Design as base model.

Quick bot creation wizard
Below shows a quick prototype to the flow of the bot creation wizard for a first time user that lands on the landing page. The wizard will end with a sign up screen.

Other Works