Overview
Pitch Protocol
Shaping an AI experience to help investors find the companies worth backing
The fundraising process was broken on both ends. Founders were pitching the same deck ten times over, chasing warm intros and reformatting materials for every fund. Investors were drowning in inbound, manually sifting through hundreds of applications to find the handful worth a conversation.
Pitch Protocol researches companies seeking funding and matches them with relevant investors. I designed the investor experience to turn that research into a clear understanding of each company, help investors examine AI recommendations, and support their next decision. Pitch Protocol supports 15+ VC fund teams reviewing 1,500+ opportunities.
- More recorded investor decisions
- 15.6%
- Pitch-to-meeting ratio
- 79.3%
- Higher filter-to-company-review conversion
- +26.7 PP
- Scope
- Product design, research, website, prototyping, systems thinking
- Team
- 1 designer (me), 1 CTO, 6 VCs
- Timeline
- 5 months
My role
As the founding designer, I shaped the investor experience and the team’s internal fund-management dashboard from concept to launch.
My work extended into implementation and release. I used AI coding tools to explore design alternatives, test interactions, and ship improvements directly in code as the product evolved.
Highlights
Understand a company at a glance
Investors can quickly understand what a company does, its key stats, and its strengths and risks. The Summary brings the recommendation upfront, with access to the detailed research when they want to learn more.
Ask AI questions and get structured answers
Investors can ask about one company or their whole deal flow. Answers take the form of tables, focused explanations, or short briefs, depending on what makes the information easier to understand and use.
Keep the evidence close
Investors can follow a finding to its evidence and sources. Confidence indicators help them judge how much support a claim has and where they need to look closer.
Find companies with AI-powered filters
Investors can describe the companies they want to find, adjust the applied filters, and save their criteria as a Dynamic View that updates as new matching applications arrive.
Learning what investors wanted to know about a company
The research gave us plenty of information to work with. The harder question was what investors needed to see first and how investors reviewed a business.
I started by talking to investors about their process. Some wanted to see revenue and traction first. Others cared more about the founders or the problem they were solving.
To understand these differences, I grouped the research into categories and ran workshops where investors:
- Ranked the information by importance.
- Chose what they wanted to see when opening a company.
- Pointed out information they felt was missing.
These conversations helped me identify what mattered to investors. I then began exploring how to bring that information together on the company page.
Finding a way to tell the company’s story
At first, I tried to create a compact version of the company report. I took the important categories, picked the most useful information from each, and explored different ways to display it.
I worked through 25+ layouts, and some versions reached production. Four approaches helped me understand why the page still wasn’t working.
The information was organized, but there was too much to read
I grouped the research into clear sections and highlighted important findings.
Investors could find individual topics more easily, but they still had to read a long report to understand the company. To give them a quicker starting point, I tried bringing more information into one screen.
Everything fit, but the company’s story got lost
I brought metrics, charts, and findings into a compact dashboard. It reduced scrolling, but the numbers didn’t explain what the company did or why it was promising.
More charts didn’t make the story easier to understand
Investors were familiar with pitch decks, so I explored charts with short explanations. But some charts took longer to understand than a sentence. I brought written context back.
The explanations helped, but the page still felt overwhelming
I grouped explanations into clear sections and kept key stats upfront. The context helped, but the page still felt overwhelming: I was trying to fit most of the report into the first view.
Request to customize the dashboard
Across these versions, investors asked for different information to appear first. One person wanted certain metrics at the top, while another wanted more space for a different part of the research.
We included customization in the early approach so people could arrange the dashboard around their needs. But that meant they had to organize the page themselves before it worked well for them.
I began to question whether another layout or more customization would solve the problem.
The breakthrough was choosing what mattered
Until then, I had been asking:
How might we turn extensive company research into an overview investors can easily understand?
After the explorations, I reframed it:
How might we give investors a trustworthy company snapshot without overwhelming them with the research behind it?
The new question changed how I chose the content. I focused on the findings that explained the company and the evidence supporting them, rather than shortening every research category.
The final approach had two layers:
- Summary: A company’s résumé, introducing the business, its key stats, and the recommendation.
- Details: The full research, explanations, evidence, and sources.
Most points in the Summary linked to the relevant research. Investors could read the snapshot first, then open a particular finding when they wanted to understand it more deeply.
The goal was to give investors roughly 80% of the information they needed for a first review in 20% of the content.
I brought the Summary together through three decisions:
- Put the recommendation and key stats first. Investors could see the assessment and essential company information before reading the explanation.
- Use one radar chart to show the overall picture. It combined scores for team, market, product, traction, competition, and risk. Investors could see the company’s strengths and gaps together instead of interpreting a separate chart for each category.
- Keep short explanations beside the numbers. The writing explained what the scores meant and preserved context that charts alone couldn’t convey.
The new Summary reduced the need for customization
Once investors had this shared starting point, they asked to rearrange the page less often. The Summary was already showing more of what they needed, so we could leave dashboard customization out of the final approach. Other teams also adopted the format.
The drop in customization requests gave me a useful sign that we had addressed the problem behind those requests. The Summary helped investors understand a company and reach a decision in under 30 minutes.
A concise recommendation still needed evidence behind it
With the Summary in place, I worked on how investors could examine the research behind a recommendation and decide whether they agreed.
I designed ways for them to:
- Preview the evidence. Read a short preview beside a claim without leaving the company page.
- Follow the reasoning. Trace a finding through the synthesis and evidence to its original sources.
- See the uncertainty. Use confidence indicators to identify findings that needed a closer look.
Investors could investigate one finding in depth while keeping the company’s wider story in view.
Helping investors find answers when they needed them
The company page made the research easier to read. As the product grew, investors also needed answers to specific questions without searching through that research themselves.
AI chat let them ask about one company or explore questions across their deal flow. My next challenge was deciding how to present those answers.
Let the question shape the answer
A long text response wouldn’t work equally well for every question. I explored different formats for three common tasks:
- Compare companies: Show relevant differences in a table.
- Investigate a company: Give a focused explanation, with visuals where they help.
- Prepare for a meeting: Bring the research into a short brief with follow-up questions.
This approach let the interface adapt to the task. Investors could compare companies side by side or read an explanation with enough context to understand the answer.
Follow the conversation beyond the first answer
I also prototyped what happened when investors asked follow-up questions:
“Show me AI companies.” → “Only those with revenue.” → “How strong are their teams?”
Working through the whole conversation helped me explore how results should change and what information each answer should carry forward.
I reused the same components, typography, spacing, and controls across response formats. Investors could receive different kinds of answers while still knowing how to read and use them.
Making specific searches easier to control and save
AI chat gave investors a flexible way to explore. Next, I worked on filters and saved Views so they could find companies with specific requirements and return to those results later.
Manual filters worked well for simple requests
I started with familiar controls for criteria such as:
Funding round: Seed · Revenue: Above $500k
Investors could select these quickly and change them directly. But some requirements depended on the research, such as whether a founder had previously scaled a business.
I needed a way to support those requests without creating a fixed filter for every possible question.
AI handled complex requests, but investors still wanted controls
I explored letting investors describe what they wanted:
“Show me seed companies with revenue above $500k and founders who have previously scaled a business.”
This made more complex searches possible. But some investors preferred changing a filter directly to writing another prompt, especially for a small adjustment. They also needed to see how AI had understood their request.
Final approach · Combining AI with visible control
I brought natural-language requests and manual filters into the same experience.
Investors could start by typing a request or selecting filters. In both cases, they could see the applied criteria beside the results.
I followed three rules:
- Show what AI applied. Make the criteria visible and editable.
- Make small changes easy. Let investors adjust a filter without writing another prompt.
- Support follow-up requests. Offer suggestions and prompts when investors wanted to explore further.
Keeping manual controls meant showing more on the screen, but it gave investors a direct way to check and change the search.
Save the companies, or save the search
Sometimes investors wanted to keep a shortlist of specific companies. Other times they wanted to keep finding new companies with the same characteristics.
I designed two types of Views:
- Static Views saved companies that investors selected and managed themselves.
- Dynamic Views saved search criteria, so new matching applications could appear automatically.
I used the same saving patterns in Inbox and AI chat. Wherever investors found something useful, they could choose whether to save the selected companies or the criteria behind the results.
Giving the team a shared place to think
Alongside finding and reviewing companies, investors needed a place to discuss them with teammates.
I brought notes and shared findings into the company page. Investors could record observations, save useful excerpts from AI responses, and build on what others had written.
This kept the team’s discussion close to the research. They could develop a shared understanding of a company while each investor kept their own decision.
From company research to a working part of the investment process
Pitch Protocol became a workspace where investors could review research, explore their deal flow, and discuss opportunities with their teams.
Investors used the recommendations and supporting research to focus their reviews and decide what to pursue. Several companies on the platform received funding.
Feedback also pointed to more regular use of the app, while plans expanded toward a founder experience.
These outcomes reflect the combined work of the platform, research, and investor experience.
More iterations don’t always solve the right problem
The biggest step forward came when I changed the question I was trying to answer. That helped me recognize customization requests as feedback about what the page wasn’t providing.
I learned to understand what people were trying to fix before adding more options.
A familiar format can still be hard to use
Pitch decks gave me a useful direction to explore, but familiarity alone didn’t make the research easier to understand.
I learned to judge a format by how clearly it explained the information and how much effort it asked of the reader.
Think about how each feature affects the rest of the product
A search could become a saved View. An AI answer could become a note. A summary finding could lead to a source.
As the product grew, I had to design these connections alongside the individual features so the whole experience continued to make sense.
Learn the domain well enough to interpret feedback
Interviews gave me a starting point. Reading company research and working through investment scenarios helped me understand the decisions behind investors’ questions.
That understanding made it easier to translate their feedback into design decisions.