Portfolio Intelligence for Data-Driven Grantmaking

The Opportunity

The Siegel Family Endowment was founded by David Siegel, co-founder and co-chairman of Two Sigma — one of the world's most sophisticated quantitative hedge funds. Two Sigma manages $60B+ using machine learning, distributed computing, and massive datasets. SFE was built on the premise that philanthropy should operate with the same rigor: "philanthropy is society's risk capital," as they put it, and it should "drive innovation by investing in local, national, and global solutions."

They deploy $25M+ annually across three focus areas — Learning, Workforce, and Infrastructure — funding organizations like Cornell Tech ($8.5M for 2025-2030), Center on Rural Innovation ($6M for AI-integrated tech talent in rural areas), Stanford d.school ($4.2M for reimagining technology-humanity relationships), Khan Academy, Scratch Foundation ($7.5M total), and Carnegie Foundation for the Advancement of Teaching. They also fund under an "Effective Philanthropy" category — grants to organizations like The GovLab that study how to make grantmaking itself better. That's the tell: SFE doesn't just want to give well, they want to systematically understand what "giving well" means.

The gap between their quantitative DNA and their grantmaking operations is the opportunity.

Proposal

The Problem Today

SFE manages a portfolio of dozens of active multi-year grants across Learning, Workforce, and Infrastructure. Their grantees range from massive institutions (Cornell Tech, Stanford) to emerging organizations (Birmingham Promise, SETDA, Modern Classrooms Project). Each grant has its own reporting cadence, its own outcome metrics, its own definition of success. The foundation publishes polished Year-End Reviews — PDFs summarizing highlights — but the underlying data that feeds those reports lives in disconnected systems.

Here's what that looks like day-to-day for a small team managing $25M+ in annual disbursements:

Grant tracking is manual and fragmented. Multi-year commitments like Cornell Tech ($8.5M over 5 years) and Center on Rural Innovation ($6M over 3 years) need milestone tracking, reporting collection, and disbursement scheduling — likely managed across spreadsheets, a CRM, and email. When the team wants to answer "which of our Learning portfolio grantees are hitting milestones on time?" they're pulling that together by hand.

Cross-portfolio pattern recognition doesn't happen. SFE funds organizations working on overlapping problems — Khan Academy and Scratch Foundation both address learning technology, Center on Rural Innovation and Birmingham Promise both tackle workforce pipelines in underserved regions. But identifying synergies, redundancies, or emerging patterns across the full portfolio requires a bird's-eye view that manual tracking can't provide.

Field intelligence is reactive. When SFE decides to make a grant in a new area — say, AI-integrated workforce development — the research process involves manual scanning of the landscape, conversations with peers, and network referrals. There's no systematic way to monitor how their focus areas are evolving, what other funders are doing in adjacent spaces, or where emerging organizations are appearing.

For a foundation founded by someone who built a $60B+ quantitative firm, there's a meaningful gap between the data infrastructure that exists and what's possible.

What We'd Build

The pitch to SFE isn't "you need AI" — it's "your founder built Two Sigma on the principle that data, systematically applied, beats intuition. Your grantmaking should work the same way." These tools bring quantitative portfolio management to philanthropic capital allocation.

Grantee Portfolio Intelligence

The foundation layer. Connect SFE's grant management data — likely a mix of Salesforce or a CRM, spreadsheets, and grantee reports — into a single portfolio view. Each active grant gets a live dashboard: disbursement schedule, reporting status, milestone progress, and key outcome metrics. Roll that up to portfolio-level views by focus area (Learning vs. Workforce vs. Infrastructure), by grant size, by stage (new vs. mid-cycle vs. final year), and by grantee type.

The real value is in the cross-portfolio analytics. Surface patterns like: "Our Learning portfolio grantees that received multi-year commitments above $1M show 2x the milestone completion rate of single-year grants." Or: "Three of our Infrastructure grantees are working on overlapping AI policy research — should we be connecting them?" This is the kind of analysis that a small team can't do manually across 30+ active grants, but that a well-structured data layer makes trivial.

Grantee Report Analysis

SFE's grantees submit progress reports — narrative documents describing what they've accomplished, what challenges they face, and what comes next. For a small team reviewing dozens of these reports per cycle, the cognitive load is significant. Build an NLP pipeline that extracts structured data from these narrative reports: key outcomes, budget utilization, stated challenges, and milestone status. Flag reports that indicate potential problems (budget overruns, leadership changes, missed milestones) and surface them for proactive follow-up.

Over time, this creates a longitudinal dataset of grantee outcomes that's far richer than what any spreadsheet captures. When it's time to produce the Year-End Review, the data is already structured and ready — no more manual compilation.

Field & Funder Intelligence Monitor

SFE operates in three fast-moving spaces where new organizations emerge, policy shifts reshape priorities, and other major funders make moves that change the landscape. Build a monitoring system that continuously tracks: new organizations appearing in SFE's focus areas, major grants from peer funders (Ford, MacArthur, Emerson Collective, Schmidt Futures) in overlapping areas, policy developments that affect grantees, and research publications relevant to Learning, Workforce, and Infrastructure.

This is particularly valuable for SFE's "Effective Philanthropy" commitment — it operationalizes the idea that good grantmaking requires systematic awareness of the field. When the Carnegie Foundation announces a new competency-based education initiative and SFE already funds them, the system flags it. When a new organization doing AI-integrated workforce training appears in rural communities — adjacent to what Center on Rural Innovation does — it surfaces for review.

Build Phases