Conversational AI

As the Principal Product Designer, I led the design and development of Warren, a conversational AI platform tailored for Private Equity (PE) firms and Investment Bankers.

Warren, a conversational AI for Private Equity and Investment Bankers, will make it easy to access financial data quickly using simple questions. This will solve problems like slow deal evaluations for Private Equity and long data-gathering times for Investment Bankers, helping both work more efficiently with real-time insights.

Problem Context

Private Equity professionals and Investment Bankers face delays in accessing and analyzing critical financial data. Private Equity (Buy-Side) teams struggle with data spread across multiple tools, making deal evaluation slow. They need quick, relevant insights to make faster investment decisions.

Investment Bankers (Sell-Side) face time-consuming data collection for client presentations. They need rapid access to deal and market information to support quick decision-making.

Both groups need a faster, more efficient way to get real-time financial insights to improve their workflow and outcomes.

Constraints

Limited Resources:
Balancing the development of both desktop and mobile features with a small team.

AI Hallucinations:
Ensuring Warren provides accurate, relevant financial data without errors.

Data Integration:
Pulling data from multiple sources while maintaining accuracy across platforms.

Mobile Complexity:
Simplifying complex financial data for mobile screens while ensuring usability.

Legacy integration:
Current platform was might not be flexible to support the demands of modern AI-driven tasks.

Goal

To modernize financial workflows by providing instant access to trusted, real-time data through Warren. Using conversational AI, users can easily access critical insights via natural language, enabling faster, more informed decisions. This shift from manual tasks to intelligent automation keeps professionals ahead in a fast-paced environment.

Team & Structure

Design Process

Research & Discovery:
• Understand our users
• Competitive analysis
• Reference data
• Prioritize with PM

Concept & User Test:
• Design Strategy
• Explore solutions
• Three concepts options
• User test and validate

Dev Handoff:
• Finalize visualize design
• Scope with PM and Devs
• Prep for development
• Provide design support

Research & Discover

Key Insights

70%

Investment bankers say that inaccurate data can cost them business opportunities

50%

Finance professionals are using AI for real-time, accurate data to reduce errors and speed up workflows.

Design Strategy

The storyboard guided this strategy which informed the design of Warren to meet the specific needs of these financial professionals, improving their workflow and decision-making efficiency:

• Simplify data access:
  Warren would provide fast, high-level financial summaries with deeper options for analysis.

• Prioritize mobile access:
  Focus on real-time updates and quick insights for investment bankers on the go.

• Deliver contextual insights:
  Use AI to offer relevant insights that help both Private Equity and Investment Bankers make quicker, informed decisions.

How might we’s:

1. What might we do to ensure the AI understands complex financial queries?

2. What might we do to balance simplicity with detailed insights?

3. What might we do to create a professional yet approachable tone for the AI?

4. What might we do to keep user trust by ensuring data accuracy?

5. What might we do to handle AI errors without disrupting the user experience?

6. What might we do to make the AI remember context during conversations?

I brainstormed features that would solve users' specific pain points:

• Natural Language Query:
  A conversational AI that responds to queries like “What’s the EBITDA of Company A for the last 5 years?” without needing users to sift through multiple tools.

• Real-Time Alerts:
  Mobile notifications for market shifts or deal changes, particularly for M&A bankers who are frequently away from their desks.

Prototyping & Testing

User Feedback

Overall Concept 2 was well received

Key Findings:

1. Conversational Approach:

   • Concept 2 was well-liked for its natural, friendly tone, making the AI feel approachable and easy to use.

2. Visuals for Quick Insights:

   • Users appreciated the visual charts and graphs, which made it faster and easier to scan and understand data.

3. Challenges with Broad Queries:

   • Users found that broad questions sometimes led to incorrect answers from the AI, which didn’t always match their intent.

4. Need for Answer Verification:

   • Users wanted a way to verify the AI's answers by seeing the data sources to ensure accuracy for reporting.

5. Establishing Trust:

   • Transparency was key. Users suggested adding a “Verify Data” button to check the source of the information and directly export verified data to reports.

Conclusion:

Concept 2 was well-received for its friendly tone and visual design, which made data easy to understand. However, broad questions sometimes led to incorrect answers, and users wanted more control over verifying the AI's responses.

Next Steps:

• Add a "Verify Data" button to cite sources, giving users confidence in the accuracy of information.

• Allow users to export verified data and visuals directly to reports.

• Improve Warren’s ability to ask clarifying questions when queries are too broad, ensuring more accurate responses.

• Test with visual designs

Visualize Design

The visual design for Warren was inspired by a unique blend of three characters and our parent company’s branding colors. Which consisted of red, black and white.

The financial wisdom and calm confidence of Warren Buffett.

The trustworthy and supportive demeanor of Alfred from Batman.

Prototype

• Check out the prototype here