B2B AI Copilot

MIGHTYBOT 

AI Copilot for Enterprises: Chrome extension designed to boost productivity for enterprises

OVERVIEW 

MightyBot is a startup developing an AI productivity copilot for revenue teams in B2B enterprises. The mission is to supercharge productivity, streamline tasks to make work easier, and bring teams together to enhance decision-making, boost sales, and help teams focus on what they do best—building relationships with customers. 

As one of the founding design team member,  my team's challenge was to execute an entirely new product from 0-1, applying design processes including concepting, user research, prototyping, user testing, and design workshops facilitation to guide product prioritization and roadmapping. 

This case study showcases our team's process in designing an AI copilot from concept to MVP.

ROLE
Director of Product Design

Led user research strategy & execution, driving data-informed product prioritization decisions. 

Headed end-to-end product design, user research & workshop facilitation, delivering investor-ready demos within 2 months. 

Directed UX research recruitment & achieved 25% response rate (vs less than 10% industry standard), establishing a research base for productive roadmap creation.

Applied human-AI UX evaluation frameworks to optimize GenAI content & prompt engineering, collaborating with engineers to improve output & user satisfaction.

PROCESS
UX/UI design, user research, user interviews, user testing, high-fidelity prototyping, design workshop facilitation, persona creation, user journey mapping. 

TIMELINE
2023–2024

CHALLENGE: BOOSTING SALES PRODUCTIVITY WITH AI

Revenue teams spend only 28% of their time selling, with the rest consumed by tedious administrative tasks like data entry and tracking sales pipelines. A significant opportunity exists to improve end-to-end sales workflow experience.

Click here to view Salesforce industry report

Industry research shows:

Streamlining tasks allows sales to spend more time with customers & close more deals.

Top performers use AI, expecting significant sales boosts.

94% of sales teams aim to consolidate their tech stack.

Leadership seeks to streamline processes for employee retention. Inefficiency has caused high turn around.

How might we leverage AI to optimize sales workflows and give teams time for what they love most—building customer relationships?

RESEARCH STRATEGY

We conducted exploratory user research to deeply understand the problem space and gather insights necessary for developing an effective AI copilot tailored to B2B enterprise sales needs.

Our strategy wasn't just about data collection; it was about gaining actionable insights to develop a tool that would seamlessly fit into and enhance existing sales processes. 

Research goals:

To identify key pain points in current sales workflows that can be effectively addressed by AI, focusing on tasks that consume time.

Uncover specific inefficiencies in sales processes

Determine which tasks are most suitable for AI automation

Understand how to free up time for relationship-building activities

METHODOLOGY

In developing our AI copilot for sales teams, we crafted a research strategy that balanced depth with practicality. We adopted the SEER framework from Sprig, (Scaling Empathy and Evidence in Research) which enabled us to gather rich user insights efficiently. This approach was particularly valuable in understanding the nuanced challenges faced by sales professionals in their daily workflows.

We also incorporated Google's UX + AI guidelines, ensuring our AI integration was thoughtful and user-centric from the start. This combination allowed us to iterate rapidly on our product concept, grounding each development phase in real user needs and expectations.

By focusing on these frameworks, we were able to create a research process that was both thorough and agile, directly informing the development of our AI copilot to address genuine pain points in the sales lifecycle.

Visualization of our research roadmap detailing the processes & methods

TIMELINE 

At the start of conception in August 2023, our team conducted the first phase of exploratory research. We intereviewed  revenue and customer success teams who are daily users of apps like Salesforce to help us better understand their workflows, user needs, and pain points.

Our team began the second phase of our research in Q1 of Fall 2023. We recruited decision-makers from revenue and customer success teams. The goal for this phase was to deeply understand the major challenges these teams face and determine what aspects of our product would be most appealing to help solve those challenges. We focused on collecting data to identify essential, high-value use cases and uncover behavior patterns that our AI agents can automate.

USER RESEARCH PROCESS

The goals of our user research were to identify the critical behavioral patterns that cause frustration for revenue and customer success teams. These insights guided us in pinpointing user needs, key categories, and vital use cases, ensuring our product development process remains human-centric and effectively reduces the risk of creating features that don't align with our users' needs.

Key learnings:

Most important to sales teams are building relationships with their customers. It’s the most enjoyable part of their jobs.

They are overwhelmed with manual & tedious tasks, & multiples tools which slows down their sales process.

Based on our research and industry reports, we've developed a sales persona to better understand and meet the specific needs of sales professionals. This persona focuses on sales representatives and account managers who play a crucial role in driving sales and revenue. By considering this archetype, we ensure that our product development is user-centered and grounded in real-world needs

This Venn diagram encapsulates our research insights, allowing us to easily visualize customer needs and frustrations. It serves as a guide in identifying essential use cases and problems to address. By reviewing the pain points and user needs, we can translate them into actionable use cases.

AI ADOPTION LEARNINGS

In our interviews, we discovered that accuracy, transparency, and user input are crucial to our users. These findings align with scholarly research on human interaction with AI technology, which has been a focus for several decades. These elements are foundational in developing a Human+AI framework to build user-centered products.

Why does it matter? 

Accuracy in AI systems is critical as it directly impacts its reliability and trustworthiness in a product.

Transparency is essential for building trust. People seek an easy way to understand AI decisions, access thorough documentation, and ensure data traceability. These elements contribute to greater trust and a more positive user experience.

User Input: While people value the convenience of machine assistance, they prefer to maintain control in the decision-making process. It’s about creating a symbiotic relationship with AI systems rather than allowing them to completely take over their work.

CROSS-FUNCTIONAL COLLABORATION: DESIGN WORKSHOP 

I facilitated a design strategy workshop and it was crucial in bridging the gap between business objectives and user needs. I carefully structured the workshop to encourage diverse perspectives, using exercises that prompted participants to think critically about our target users' pain points and potential AI solutions.

The workshop's outcomes directly informed our product roadmap, allowing us to prioritize development efforts on features that would address the most pressing needs of enterprise sales teams while aligning with our business goals. This strategic approach set a clear direction for our MVP, ensuring we were building not just an innovative product, but one that would truly resonate with our target market.

Below is a snapshot of the user journey and product strategy mapping we accomplished in the workshop. 

Completed user journey map of a sales lifecycle. 

This exercise uncovered opportunities and high-value tasks to automate, and influenced our product prioritization & roadmapping.

UNCOVERING CUSTOMER USE CASES

Our research discoveries revealed high-value AI use cases for sales by identifying critical behavioral patterns that frustrate revenue teams. These insights enabled us to pinpoint real user needs and key categories, ensuring a human-centric product development process.

The resulting AI use cases include personalized prospecting emails, comprehensive meeting preparation, post-meeting follow-ups, integrated deal management, data-driven forecasting, and automated win/loss analysis, all designed to align with user needs and enhance efficiency.

“This is very cool. This solves a lot of problem of existing AI tools. Imagining how I use ChatGPT and our proprietary AI systems, there's still so much manual work that needs to happen. This just kills all the extra, unnecessary work which is awesome.”
– Addie L., Head of CX at tech startup

UI/UX DESIGN

CHAT FEATURE
Balancing human agency & machine capabilities

In designing our AI copilot MVP, we applied key design principles to optimize user experience: prioritizing simplicity and clarity, using conversational language and tone, and providing simple explanations of how the AI system works to build user trust. 

We focused on concise messaging, incorporated user feedback loops for system adaptability, and maintained consistent, familiar design conventions for chat interactions. Additionally, we supported multimodal inputs and outputs to enhance the overall user experience.

AUTOPILOT FEATURE
Simplify manual workflows

When designing AI automation for our MVP, our key considerations included enabling users to create Autopilots that simplify workflows through human and machine co-learning. This approach evolves with user input to improve performance.

We ensure concise notifications that inform users why they're receiving them and provide proactive assistance. Capturing user preferences through proactive interactions allows users to supervise and maintain the decision-making process. Gradually, the Autopilot system can take on more responsibilities under user guidance.

ONBOARDING JOURNEY
First impressions matters

It is critical that the first interactions a user has is positive. It can set the tone for the user’s entire relationship with the product, which can lead to increased adoption, satisfaction, and loyalty.

I facilitated a UX workshop with our team to explore a user's journey and map what a successful onboarding experience for MightyBot would look like.

The workshop centered around the challenge: "How might we achieve a successful onboarding experience that builds trust with the user?" Our goal was to create a user-centered onboarding process that fosters trust and engagement. By focusing on positive and valuable interactions, we aimed to drive user adoption and ensure their success with the extension.

Below is a snapshot of the user journey map we accomplished in the workshop. 

USER TEST RESULTS FOR ONBOARDING PROTOTYPE

After our ideation workshop, our Senior UI Designer created a high-fidelty prototype of our onboarding experience and tab features for user testing. 

Our team conducted test and gathered valuable feedback and insights. By observing real users interact with the product, we identified areas where the experience could be enhanced and where potential issues might arise. These insights were crucial in informing our product iteration and improvement, ensuring that our MVP is closely aligned with user needs and expectations. 

User Test Objectives:

Test our design solution to evaluate usability and efficacy.

Identify problems in the design of the experience.

Uncover opportunities for improvement.

WELCOME EMAIL

This is the first communication a new customer receives when onboarding to a software product. It sets the tone for the user’s journey by providing a warm introduction and essential information to get started. This email typically includes a brief overview of the product's features, helpful resources or tutorials, and contact information for support. The goal is to make users feel welcomed, valued, and confident in using the product, ensuring a smooth and engaging start to their experience.

User feedback:
Too much text.

Users just want to see simple a explanation.

They just wanted to click on button to get immediately started.

Users felt they didn't need complex instructions.

They want to see more visuals such a gif animation to show how product works.

ACTIVATION

The activation page for MightyBot Chrome extension should provide clear and concise instructions to help users seamlessly integrate the extension into their browser. This page guides users through the essential steps, from installing the extension to enabling its key features. With easy-to-follow visuals, our aim is to provide a hassle-free setup experience, allowing users to immediately benefit from MightyBot's capabilities.

User feedback: 
Users want to see help center link.

Make text less wordy. Simplify messages.

Product descriptions on this page is unclear.

Users prefer to see more information about how to use producs.

Users like the gif animation and info on how to pin extension to their browser.

Most users liked the “Start Here” graphics but a few also said they didn’t need to see it.

CONNECTING APPS

Designed to make linking user's software quick and effortless, the connect app page provides a straightforward interface where users can easily integrate their existing software with MightyBot. With just a few simple steps, users can connect their apps and immediately begin leveraging the powerful features of MightyBot. Our goal is to streamline the connection process so users can instantly start automating tasks and enhancing productivity without any technical hassle.

User feedback:

In the extension, it wasn't clear to users there were 3 separate steps to get started. They wanted to see it distinctly labeled as Step 1, Step 2, Step 3. 

Connecting apps were intuitive and easy to do.

After connecting app, users want to immediately start using chat and not go through all the steps to set up their Autopilot.

AUTOPILOT TAB

MightyBot's Autopilot streamlines your workflow by automating repetitive tasks, providing quick access to needed information, and offering smart analysis of live deals to boost your focus on likely successes. It also delivers real-time customer insights and coaching to enhance your sales strategies. Integrating seamlessly with tools like Salesforce, Google, and LinkedIn, Autopilot maximizes your productivity and efficiency.

User feedback: 
Users were overwhelmed with all the content. They prefer to see 1 example Autopilot to get started. 

Users liked that the Autopilot tasks were highlighted in green for easy identification. 

When clicking on each Autopilot task, users expectated the out put text to clearly describe what it's doing and also   pre-determine or suggest what they need to do next to succesfully complete that task . 

Users appreciated the “What’s this” help link to guide them through the features in Autopilot.

Users were confused by the term “Interval”. It wasn't clear to them that it was label to describe their Autopilot setting. 

HOME TAB: DAY AT A GLANCE FEATURE

Home tab in MightyBot is the usere's daily dashboard, displaying both their preset Autopilot tasks and those prioritized by MightyBot based on their schedule. It helps them focus on key tasks and enhances their productivity throughout the day.

User feedback: 
50% understood what this feature is suppose to do, but 50% were also confused.

Users want MightyBot to tell them what tasks they need to prioritize instead of just showing them what tasks they have. They want it to truly assist them on next best action to take to be successful in their day. 

Transparency: users want clear explanation on what MightyBot will do for each of their tasks.

Users didn't like that the output contained too much text in the chat. 

User liked the thumbs up/down feedback icon. They want some decision-making and oversight on what interactions with MighyBot worked or didn't work. 

AUTOPILOT TASK NOTIFICATION:
CRAFTING EMAIL RESPONSE EXAMPLE

Here's an example of an Autopilot notification on the Home tab. MightyBot has automatically drafted a response to a sales lead based on previous emails. When user clicks on the task, it opens the draft in Gmail for their review.

User feedback: 
Users were concerned that MightyBot would just send the email. They want make sure they can review all email drafts before sending to ensure it's accurate. 

Some users want to see “Let’s craft a response” as the link instead of “targeted leads”.

Users said that the tasks that MightyBot prioritizes automatically must be relevant to their needs or they won’t use it.

Some users didn’t want to keep seeing the suggested prompts above the chat box. they want to see it once when the log in and then have it disappear after they have interacted with the chat. 

ONBOARDING PROTOTYPE ITERATION

Based on the feedback from our first user testing, our design team refined the onboarding flow as well as the features for each tab based on the insights we learned from user feedback. This was critical to help us make informed decisions that are based on real user data and not on our team's assumptions on what we think user's might want. 

ADDITIONAL USER FEEDBACK 

Majority of users rated MightyBot highly. Our average rating was 8 out of scale from 1–10. 

Users validated our assumptions that Autopilot feature does improve their productivity, but the automated tasks must be relevant to their workflow or they will ignore it. 

Users see potential applications beyond sales activities and wants to integrate extension into other parts of their workflows.

People are concerned about privacy, data security, accuracy and response time.

They want the extension to integrate with more apps in their daily tool use. 

Users want the ability to customize and turn off features so they have oversight on what they want MightyBot assist with. 

Results from our System Usability Scale (SUS) survey. 

OUTCOME + REFLECTION

As a found design team member, Director of Product Design, at MightyBot, leading the development of an AI copilot was a transformative experience that honed my ability to leverage human-AI principles to guide product development, and learned the value of elevating research insights to inform not just product decisions, but also broader business strategy. By consistently translating user needs into potential market opportunities, we were able to position our AI copilot as a key differentiator in the enterprise software space. This experience reinforced the importance of a design leader's role in bridging user-centered design with business objectives, ultimately driving innovation and growth.

Human + AI Framework Considerations 
By adopting a human-centered approach to AI product development, we can not only boost key aspects of user experience but also amplify successful AI integration.

The path forward lies in symbiotic partnerships between humans and AI. With humans having agency in their expertise, creativity, and judgment in collaboration with AI power, we can pave the way for a harmonious human-AI experiences that emphasizes human growth and job satisfaction.

As product designers and UX researchers, it's critical for us to focus on the strengths of both humans and machines, using human judgment, creativity, and adaptability alongside the computational power and speed of AI. 

While we see that people appreciate the convenience of machines assisting them, they still prefer to have control in the decision-making process. It’s about building symbiotic relationship with AI systems rather than having it completely take over their work.

To practically and effectively measure and optimize an AI-powered user flow it is important to implement a “user-centered evaluation framework.” This framework allows product teams to quantitatively measure the impact and efficacy of AI in driving better results throughout the user journey.

Additionally, this experience reinforced the critical role of research in product design. Conducting research can facilitate collaborative efforts between developers and designers and researchers. It's a holistic approach to software development, where design decisions are informed by technical possibilities and limitations, and engineering solutions are guided by design principles and user needs.

Human emotions have profound influence on our interactions with digital and physical interfaces. Attractive, fun, and enjoyable products are more engaging and effective than purely functional ones. Emotions are a key driver of human behavior and decision-making.

MEDIUM ARTICLE

I co-authored an article with Douglas Melchoir, formerly VP of Product at Ocrolus where we shared our insights and learnings in developing AI/ML SaaS products for enterprises. 

AI systems should leverage user experience design, applying frameworks that focus on the strengths of both humans and machines, using human judgment, creativity, and adaptability alongside the computational power and speed of AI.

Link to Medium Article