When looking at leading startups in the Midwest, you'll likely find Drive Capital as an underlying catalyst behind their success. With offices in the US and Canada, the $2.2B venture capital firm strategically invests in world-changing companies spanning healthcare, insurance, finance, robotics, and education.

Since 2014, Drive has been building its own AI-powered CRM called Herbie to source investments and win funding deals. I worked with the engineering team to make an investor sourcing tool called "Market Maps" more user-friendly by bringing in the underlying machine learning to accelerate procurement.

• Improved Investor Efficiency: With new Market Map tools and enhancements, 16% of all investment white papers are now derived from companies recommended by Herbie's AI.
• Continuous Feature Updates: The Market Map improvements are in development, with new search and filter features already integrated in response to evolving needs.
• New Emphasis on Investor Experience: Drive is fundraising this year and qualitative feedback from their limited partners has skewed positive, with new initiatives shaping our roadmap.

To kickoff our collaboration, we met with Drive's engineers and founders to learn about how investors used Market Maps and their pain points. Following a deeper usability audit of the tool, we identified several problems to solve:

1. Convoluted Interactions: Market Maps were static without useful interactions like drag-and-drop and filtering.
2. Fixed Views: The experience only worked on large desktops, making it difficult for investors to use from their mobile phones.
3. Lack of Actionability: Market Maps didn’t support features to highlight specific segments or collect insights about specific companies.
4. Missing AI Intelligence: Investors had to manually find, add, and sort companies in their Market Map and could not use Herbie's machine learning capabilities to find relevant companies.

Stressing the importance of iteration over innovation, the engineers wanted to slowly introduce Herbie’s AI capabilities into Market Maps while layering UX/UI changes over the experience.

I facilitated a "keep, add, improve, remove" brainstorming and prioritization exercise to clarify Drive's goals, align on new features, and identify specific technical considerations. I translated decisions into a requirements document, recording user needs and business criteria before sharing with the engineers for visibility and alignment.

We started ideating on early concepts, communicating new interface patterns and AI features with wireframes to align on updates before screen design. UX/UI improvements included:

• Canvas and Table Views: An improved "canvas" and a new "table" provide complimentary views of detailed company information.
• Company Panel: A proposed “contextual panel” surfaces company metadata and contact events with Drive to help users get information quickly and easily.
• Segment Analysis: Machine learning analyzes user-defined segments to suggest similar companies and streamline the procurement process.
• Smart Search: A new search feature interprets natural language to help investors explore their ideas faster. For example, a search like, "Truck-bed robotics companies based in the Midwest," yields a list of related companies.

The engineers struggled with memorizing styling conventions, so before moving into screen design, we created a design system leveraging utility classes and design tokens to increase engineering efficiency. Around this time, Figma released variables that enabled closer parity between design and engineering to create and manage tokens.

I drew inspiration from Drive's website to create the UI style, opting for a minimalist design language. We knew that the investors loved using Notion for personal tasks, so we emulated some of its design language with subtle shades of neutral colors and intuitive inputs.

The foundational design system enabled our team to quickly transition our wireframes into high-fidelity screen designs. We referenced real Market Maps to design for different degrees of complexity while creating annotations and pattern documentation.

The complexity of the redesign made it difficult to track tasks and find iterations, so I streamlined the designs with a dedicated table of contents in our Figma file. The directory helped improve communication and handoff with Drive in later weeks when the engineers started implementation.

We recruited two investors for usability testing on core workflow tasks. We facilitated a moderated test using a prototype where they were tasked to create a new Market Map, reorganize companies into new segments, and interact with the contextual company panel.

• The investors relied on searching to find companies and struggled to locate it in our designs.
• Our participants responded positively to the new table view, but the utility of this feature was not immediately obvious until they independently explored it.
• They were enthusiastic about introducing the AI into Market Maps, but also hesitant to completely embrace its proposed implementation. They wanted the ability to reject recommendations from the system.
• They were intrigued by the contextual company panel, but gave us recommendations for what information and actions should be elevated above the fold.

Responding to the investors' feedback before handoff to the engineers, we spent the last phase making improvements to the navigation, search, and company panel. Key contributions I made to refining our solution include:

• Rethinking the design and UX of the canvas navigation, moving the search into a more prominent location to improve discoverability.
• Introducing interactions to remove recommended companies on the canvas, table, and left-side "market" panel.
• Eliminating and reorganizing information on the company panel and redesigned CTAs on groupings of information to go deeper.

Herbie’s machine learning is now imbedded in the search, allowing users to find similar companies with semantically-labeled indicators showing how strongly results correlate to other companies in their Maps. Users can also explore their ideas with natural language prompts in addition to traditional queries.

The system assists in procurement by analyzing the nuanced ways users structure their maps to suggest relevant companies in niche segments. Investors can always hide or reject Herbie’s recommendations from their Maps with the click of a toggle, ensuring that procurement always keeps the human as the driver of the tool.

The new Market Maps experience features four primary elements: the Canvas, the Table, the Market Panel, and the Company Panel.

• Most of the screen is dedicated to the "map," from which users can toggle between the grid-based Canvas view or cell-based Table view.
• The Market Panel offers users helpful tools such as search, map insights, categories, and recently viewed companies.
• Interacting with individual companies opens the Company Panel showing relevant information like founder contact information and business insights.

Market Maps have been redesigned for full responsiveness, ensuring that investors can access key information anywhere, whether they’re working from the Drive offices or taking an Uber to the airport to meet with founders.

By referencing real Maps from investors, I tailored the responsive patterns to handle various levels of structural complexity to accommodate individual investor preferences.

Inspired by Notion, a tool that Drive’s investors love, Herbie’s new UI emphasizes minimalism and discrete interactions in staunch comparison to the old information-dense experience.

The UI has been extended into other parts of the platform such as company profile pages, improving engineer efficiency with a reusable library of components and foundational styles.