SURGE AHEAD

Unifying Cohort Data With AI

Newly funded founders often have access to capital, but not clarity. Across countries, industries, and funding stages, cohort members struggle to find the right resources, answers, and people at the right time. I explored how we can centralize knowledge and community to help founders move faster.

Sequoia’s online network, SURGE, was designed to rival Y Combinator’s Bookface by supporting early-stage founders beyond capital. However, the core experience was fragmented: critical information lived across multiple platforms, channels, and repositories making it difficult for founders to discover value when they needed it most.

Machine Learning
Platform Design

Role & Scope: I was responsible for improving the usability and value delivery of the SURGE platform, with a focus on strengthening the cohort experience and making community impact more measurable. Partnering closely with the Community Manager and full-stack engineers, together – we focused on translating qualitative community insights into product decisions that reinforced engagement, clarity, and long-term value.

“Providing value to the user is the clearest signal for where innovation should happen next.”

Approach – Framing the Problem

Before designing solutions, we needed clarity on who the platform was for and what success looked like for them. I used the Community Canvas framework (a tool for designing purpose-driven communities) and aligned it with core product design principles to ground decisions in user needs rather than features. This helped shift the conversation from “What should we build?” to “What problem are we solving for founders?”

Identified User Purposes

  • Gather with Community: Founders need a trusted space to learn from peers and avoid building in isolation.
  • Find Relevant Answers: Each startup faces unique challenges. Founders need fast, contextual access to insights that match their stage and situation.
  • High-Value Resources: The platform should reduce friction to exclusive tools, perks, and guidance – helping founders get to market faster.

Evaluation – Does SURGE Deliver Today?

  • Video Academy
    • Provides structured value, but discovery and relevance could be improved.
  • Resources & Perks
    • High-quality offerings, yet fragmented across multiple entry points.
  • Community Channels (Connect, Messaging, Feed)
    • Active communication exists, but conversations and insights remain siloed and difficult to reuse or measure.

Key insight:
Value existed—but it wasn’t centralized, contextual, or easy to surface at the moment founders needed it.

The Research

It’s easier to mass message on the WhatsApp group to get an answer
I use search mechanism on Notion, don’t have time really to parse through the notes
I messaged a founder from a previous cohort with a similar company strategy to mine, but didn’t hear back

We interviewed the current cohort.

Got some common sentiments from the founders regarding their feedback on the current design and what they actually want to accomplish.

75

%

Of users were coming for Videos, making E-Learning the largest lead-gen on platform. But it wasn’t sticky enough for founders to keep coming back to the platform.

25

%

Of remaining users came to Claim a Perk, and there was a large drop off after.

The Messages feature was least used due to low integrity on use-case, but most desired.

Results:

Founders want to:

  • Connect to previous founders for guidance
  • Find answers faster to experience-related questions
  • Feel Close Knit to other founders and feel supported by the community

Redesign

We took this user research and took some actionable items to redesign the platform to meet closer to the needs of the user. The methodology was to take existing user behaviors such as messaging on WhatsApp groups, and rather than fighting it, using it to our advantage to create a stronger platform.

“Tackling the largest problem was to unify all the data across the different channels into one succinct location.”

The idea was to train an AI model with all the data and resources (including the WhatsApp group archives) and take a Question-Answering Over Documents approach via LLM to learn the different ways to successfully answer the founder, some options were Question-Document based Retrieval for accuracy.

The most important aspect to this was making sure to include a recursive feedback loop which concluded whether the search itself was actually helpful or not. (Using deep learning techniques and not actually knowing whether they’re helpful or not can tend to be more frustrating than actually helpful to the user.) Positioned in the same location as a news feed post, since that’s where the heatmaps were indicating where users where heading to most on the home page.

More Work

View All
Sequoia Crypto Explorer

Sequoia Crypto Explorer

Bijlee Web3 Grocery

Bijlee Web3 Grocery