The Benevolent Bandwidth Foundation

AI for Humanity

We believe the most valuable resource in the AI age isn't compute or data, it's our Benevolent Bandwidth. It's the human capacity to care, build, and ensure technology serves the many instead of the few.

Our Mission

We solve problems with transparent and open technologies that deliver public benefit by prioritizing safety, fairness, and trust.

Our Vision

We are building a world where the benefits of AI are shared by everyone, strengthening fairness, opportunity, and human agency.

How We Create Impact

1

Identify Problems

We identify important problems and ideas that deliver public benefit which can be solved with technology

2

Onboard Contributors

We identify, motivate and onboard technical advisors and contributors

3

Connect Ideas

We connect ideas with contributors

4

Execute

We execute on the ideas and connect users with the tools

The b² Principles

Our Organizational Formation

1

Foundation Advisors

Stewards of the foundation

2

Technical Advisors

Owners of projects and repos

3

Benevolent Contributors

Execution leaders

Role of Foundation Advisors

Strategic Oversight & Guidance

  • Review the project portfolio
  • Steer the health/engagement/impact of the foundation
  • Ensure the organization is living by its mission, vision, and principles

Connection & Growth

  • Identify new ideas
  • Connect the foundation with potential contributors at scale
  • Help open doors and unblock the organization

Member Engagement

  • Ensure all members are engaged and benefiting from their participation

Role of Technical Advisors

Benevolent Foundation Advisors

Dan Loehr

Dan Loehr

Associate Professor of AI in Climate Change, Georgetown University

Previously: VP of Data Science, Capital One

Nicole de Picciotto

Nicole de Picciotto

Founder, NdP coaching

Previously: VP of Community Investing and Impact, Capital One. COO, DataKind

Koen Powels

Koen Powels

Distinguished Professor of Marketing and Assoc Dean of Research, Northeastern University

Shana Sood

Shana Sood

Chief Marketing and Communications Officer, Prudential Financial

Previously: Head of Core Mrketing, CapOne

Yi-Chen Tu

Yi-Chen Tu

Managing VP of Data and Data Science, Capital One

Previously: Dir of Research, Travelers

Aleksandar (Sasha) Tomic

Aleksandar (Sasha) Tomic

Associate Dean and Program Director for Economics and Analytics, Boston College

Nurtekin Savas

Nurtekin Savas

Founder, The Benevolent Bandwidth Foundation

Head of Global Credit Infrastructure, Data and AI, PayPal

Gabriel Safar

Gabriel Safar

Head of Real Estate Solutions, Proxet. Managing Principal, Line Advisors. Attorney at Law

Previously: Co-Founder and CEO, LeasePilot

Benevolent Technical Advisors

Kal Mishra

Kal Mishra

Director of AI, Capital One

Previously: Fermilab, Stanford Linear Accelerator Center

Debalina Gupta

Debalina Gupta

VP of Data Science, ANZ Bank

Previously: Fidelity Investments. Director of AI

Rio Cruz

Rio Cruz

AI Product Founder, HappyFin

Previously: Velocity Black and Capital One

Isha Chaturvedi

Isha Chaturvedi

Applied Scientist, Amazon AGI

Previously: Lead Machine Learning Scientist, Applied Research (LLMs), AI Foundations Team, Capital One

Ankush Chopra

Ankush Chopra

Senior Director of AI, Tredence

Previously: Sr Lead of AI,Fidelity Investments

Mark Boenigk

Mark Boenigk

Data Engineer and Analyst, Capital One

Previously: Data Analyst, Fragport AG

Atul Bhagat

Atul Bhagat

Member of Technical Staff, Nile

Previously: Data Science Manager, SiriusAI

Alexandra Matei

Alexandra Matei

Manager, Data Analytics & Engineering, Tesla

Previously: Data Scientist, eCare Services

Shan Ali Shah Sayed

Shan Ali Shah Sayed

BI Developer, Qvinci Software

Previously: Business Analyst, IBM. Senior Consultant, Systems Limited

Larry Fulton

Larry Fulton

Professor of Applied Analytics, Boston College

Alfy Samuel

Alfy Samuel

Director of AI, Capital One

Previously: Director of AI, Fidelity Investments

Pranay Tiwari

Pranay Tiwari

Vice President, Data Science, Fidelity Investments

Previously: Sr. Research Manager, IMRB International

Our Approach & Focus Areas

Our projects represent a vision for using AI and technology to solve real-world problems. Each idea has been carefully vetted to ensure alignment with principles of public benefit, privacy protection, and open-source development. Active projects and our backlog span several key problem domains where technology can create meaningful impact. Our goal is to develop deep expertise in each domain over time, building specialized solutions that address systemic challenges through AI-powered tools.

Our initial focus areas span the following domains:

Over time, we aim to partner with non-profits, NGOs, and other public benefit organizations to jointly solve problems at scale using technology and AI, amplifying the collective impact on the communities we serve.

Active Projects

Research Agent

Problem domain: Access to Information

Too many important papers get lost in the noise. Most researchers and practitioners cannot reliably scan what is new recently in their area, find truly promising work, and trust that they did not miss something big. This agent helps with this problem by finding, ranking, and explaining recent AI papers on arxiv.org.

Try It Now →
"I love the way the agent works. I will keep it in my toolkit"
— Google Engineer
"I recommend the agent to my clients"
— AI in Marketing Expert
"This is a great use case for an agent that actually earns the name"
— A Reddit reviewer

EcoTag: The Garment Emissions Scanner

Problem domain: Environmental Impact

EcoTag is a smartphone app that empowers consumers to make sustainable choices by revealing the hidden environmental cost of their clothing. Using Optical Character Recognition (OCR) to scan garment tags, the tool calculates estimated lifecycle carbon emissions based on materials and origin. This provides shoppers with immediate, actionable data on the carbon footprint of their potential purchases.

Delivery Optimizer

Problem domain: Logistics

The Delivery Optimizer helps local businesses like florists and dry cleaners eliminate the waste of manual route planning. Owners simply upload a list of addresses to the web tool, which uses open-source algorithms to generate the most efficient delivery order. This streamlined process saves time and fuel while reducing the carbon footprint of local logistics.

The Revive-or-Recycle Scanner

Problem domain: Environmental Impact

People don't know if fixing old gadgets is worth the cost, so e-waste accumulates. This economic triage tool queries live marketplaces for part and resale prices to advise consumers whether to repair or recycle their old or broken electronics, helping reduce environmental waste while saving money.

The Ultra-Processed Detective

Problem domain: Nutrition

Healthy marketing is deceptive and identifying ultra-processed foods on the NOVA scale is hard for consumers. This red light/green light scanner helps grocery shoppers avoid ultra-processed foods by using OCR and AI to read ingredients and classify foods, providing instant guidance on food quality.

Second Look

Problem domain: Healthcare

Screening mammography has limitations, including false negatives and false positives. Individuals often lack a meaningful "second signal" on their results, or they experience anxiety from indeterminate findings. Second Look is a consumer-facing, AI-assisted interpretation companion that analyzes user-uploaded mammography images to identify regions that warrant attention and assign a coarse concern tier. It provides informed, conservative decision support to prompt professional follow-up, not to replace a doctor. The tool utilizes on-device inference for privacy and is available as a web or mobile application.

Beacon

Problem domain: Access to Information

Beacon is a project designed to protect the elderly from malicious sites, emails, and links, including AI-generated content. This tool helps vulnerable users navigate the digital landscape safely by identifying and warning about potentially harmful content, reducing the risk of scams and phishing attacks that disproportionately target older adults.

The Medical Bill Defender

Problem domain: Healthcare

Medical bills are riddled with errors and insurance denials are confusing, making it exhausting for patients to fight back. The Medical Bill Defender is a local-first, privacy-centric agent that helps patients in the US analyze their medical bills and Explanation of Benefits (EOB) documents. Users upload bill or EOB images or PDFs, and the tool checks CPT codes, flags errors, and automatically drafts appeal letters. The agent empowers patients to challenge incorrect charges and denied claims without the complexity that typically prevents them from taking action.

View Backlog View on GitHub

Get Involved

Join our community of contributors building AI tools for public good. Whether you're a developer, researcher, or enthusiast, there's a place for you.

Email Us

Get in touch with questions or ideas

benevolent@benevolentbandwidth.org

GitHub

Contribute code and collaborate

github.com/benevolentbandwidth

Slack Community

Join the conversation

benevolentbandwidth.slack.com

Collaboration Interest

Express your interest in collaborating

Fill out the form

Submit an Idea

Share your project ideas with us

Submit your idea

Why Join

Learn about benefits for contributors and advisors

Why Join