Enhancing sales team productivity through AI document analysis with Azure OpenAI integration
Company
Avanade, a global consulting firm focused on innovative digital transformation solutions for enterprise clients.
The mission
Create an AI-powered document assistant to elevate sales productivity.
Final prototype
My responsibilities
AI trends research
Market research
Prototyping
Data mapping diagrams
Cross-functional team alignment
Team
1 Project Lead
2 Developers
1 Business analysis
1 Visual Designer
Time
Jun 2023 - Aug 2023
The Challenge
Time-consuming document searches reduced sales productivity
The goal was to enable fast access to work documents and provide insights to improve efficiency.
Solution
AI assistant increases efficiency and shortens sales cycle
I led the design of an AI document assistant that reduced document search time by 70%, shortened the sales cycle by 45%, and boosted tool adoption by 35%.

By integrating Azure OpenAI, we enhanced document access and insights, allowing sales reps to be more productive and confident during client meetings.
Approach
Agile workflow to iterate and optimize
I worked closely with developers, business analysts, and stakeholders to ensure the design aligned with technical constraints and user needs.

We followed an Agile workflow, iterating on key features to optimize speed and usability across devices. I focused on centralizing the document upload process and introducing dual upload methods, enhancing flexibility for users.
Table of Content
Initial Demo
Identifying gaps in the initial prototype
"Here’s what the current demo currently looks like."
- Project Manager
My Observations
The Challenge
Sales representatives were spending up to 3 hours daily searching for and analyzing documents.
This time-consuming process decreased their productivity and eroded confidence during client meetings. They needed a quick way to analyze and retrieve information from proposals and product data.
Key User Demands
Users want a tool that is easy to use, saves time, and enhances their ability to access and present information
Problem Statement
How might we…
Enable sales representatives to quickly access and analyze critical document information to enhance their productivity and effectiveness during client interactions?
Research - Qualitative
Users addressed excessive time spent searching documents and difficulty accessing information
We conducted workshops with sales representatives, business analysts, and engineers to identify key pain points.
"I often spend hours every day sifting through documents to find the information I need for meetings. It's exhausting and takes away from the time I could be spending with clients."
- Sales Representative
Defining success
How do we measure success
We focused on three key metrics:

- 25%

reduction of the cost of the staff

- 45%

shorten insales cycle

+ 35%

increase in adoption through increasing the client trust and data privacy

Design focus
Design focus areas
Feature Prioritization
Prioritizing features to focus on speed and accuracy in document retrieval
With our sales-focused features outlined, we used the Eisenhower Matrix to prioritize based on urgency and importance.
Design iterations
Maintaining an Open Feedback Loop
Throughout the design process, we maintained an open feedback loop. I regularly shared wireframes and prototypes with both the engineering team and PMs to gather their input.
Fast Agile Cycle and Engineering Constraints
Overcoming Engineering Constraints with Flexible Design
During the development phase, the engineering team identified technical constraints related to the chatbot’s ability to handle multiple file uploads simultaneously.

Recognizing the potential impact on the project timelines, I took the initiative to redesign the user flow to address these limitations.
Iteration 1
Centralizing primary action to prompt user
The file upload feature was minimally visible, occupying only 30% of the page. To prioritize user convenience, I centralized the upload area, making it more prominent and easier to access.
Iteration 2
Integrating dual upload methods
I introduced dual upload methods—local and database uploads—so that reps could easily pull in the most relevant documents from different sources without disrupting their workflow.
Final Version
AI document assistant to elevate sales productivity
"The AI-powered document assistant has transformed the way I prepare for meetings. I used to spend hours digging through proposals, but now I can find exactly what I need in minutes."
- Sales Representative
Final Version
Rapidly search and retrieve meeting notes, proposals, and presentations
User uploads the necessary document, previews it, and prepares for the meeting.
Access tailored AI insights with relevant info and guiding questions
The AI assistant provides insights tailored to the file, helping users with relevant information and guiding questions.
Revisit chat history and files for clarification
User feels fully prepared, with the ability to revisit their chat history and uploaded files for clarification.
Visual design
Implementing responsive design for accessibility
Recognizing that sales reps are frequently on the move, I implemented five breakpoints, supporting the sales team's need for flexible document access.

At each breakpoint, I adjusted:
1. Layout
2. Functionality
3. Interactive elements
Data mapping diagram
User and data interaction

I have created a data mapping diagram that outlines the user flow when interacting with the product and details where the data will be stored. This diagram is essential for engineers during development and invaluable for new users to understand the data interaction process.

(Click to enlarge)

Reflection
The value of cross-functional communication and effective prioritization

A key takeaway from this project is the importance of collaboration and clear communication. Working with team members from various disciplines, I learned that consistent updates and sharing progress ensured alignment across the team. This transparency helped us address issues early and stay on track with our MVP deliverables.

Another lesson was learning to prioritize under time constraints. Through this project, I came to understand that it's not always feasible to meet every ambitious goal. Instead, focusing on the most impactful and essential features within the available time frame is key to delivering a high-quality product. By staying adaptable, we adjusted expectations without sacrificing quality, delivering a product that met both user and stakeholder needs.