Moroccan Founder Secures $4.2M for AI Search Startup

The Challenge of Data Retrieval in Generative AI

As generative AI continues to transform various sectors, a crucial, often unseen, obstacle emerges: retrieval. This refers to the process of efficiently accessing the correct data, complete with pertinent context, from complex knowledge bases.

The accuracy of large language models (LLMs) is fundamentally dependent on their ability to retrieve information effectively.

ZeroEntropy: A New Approach to AI Retrieval

ZeroEntropy is a San Francisco-based startup aiming to address this challenge. Co-founded by Ghita Houir Alami and Nicholas Pipitone, the company recently secured $4.2 million in seed funding.

This funding will be used to enhance the speed, accuracy, and scalability of data retrieval for AI models.

Investment and Backers

The seed round was spearheaded by Initialized Capital, with additional investment from Y Combinator, Transpose Platform, 22 Ventures, a16z Scout, and a diverse group of angel investors.

These investors include individuals with experience at leading AI companies like OpenAI and Hugging Face.

Positioning in the RAG Landscape

ZeroEntropy is part of a growing trend of infrastructure companies focused on retrieval-augmented generation (RAG). RAG is becoming increasingly important for powering the next generation of AI agents.

Competitors in this space include VoyageAI from MongoDB and fellow Y Combinator startup, Sid.ai.

Why Initialized Capital Invested

“We’ve evaluated numerous teams working on RAG solutions, and the models developed by Ghita and Nicolas consistently outperform everything we’ve observed,” states Zoe Perret, a partner at Initialized Capital.

Retrieval is a critical component for advancing AI, and ZeroEntropy is dedicated to building a robust solution.

The Power of Retrieval-Augmented Generation

Retrieval-augmented generation (RAG) utilizes data from external sources and has become a standard architecture for AI agents.

Applications range from chatbots accessing HR policies to legal assistants referencing case law.

Addressing the Fragility of Current RAG Systems

The founders of ZeroEntropy identify a weakness in many current RAG implementations: a reliance on a patchwork of vector databases, keyword searches, and re-ranking models.

ZeroEntropy provides an API that streamlines ingestion, indexing, re-ranking, and evaluation processes.

A Developer-Focused Tool

Unlike search products designed for end-users, such as Glean, ZeroEntropy is specifically designed for developers.

It efficiently processes data, even from complex internal documents, and Houir Alami describes it as a “Supabase for search.”

Simplifying Search Infrastructure

“Currently, many teams are either manually integrating disparate tools or feeding entire knowledge bases directly into LLM context windows,” explains Houir Alami.

“The former is time-intensive, while the latter can lead to errors. We are creating a developer-first search infrastructure – akin to a Supabase for search – to facilitate the deployment of accurate and rapid retrieval systems.”

Introducing ze-rank-1

The core of ZeroEntropy’s technology is its proprietary re-ranker, ze-rank-1.

The company asserts that ze-rank-1 surpasses comparable models from Cohere and Salesforce in both public and private retrieval benchmarks, ensuring the most relevant information is prioritized.

Early Adoption and Customer Base

Over ten early-stage companies developing AI agents in sectors like healthcare, law, customer support, and sales are already utilizing ZeroEntropy.

The Founder’s Journey

Ghita Houir Alami was born and raised in Morocco before moving to France at 17 to pursue engineering studies at École Polytechnique.

It was there she developed a passion for machine learning.

Education and Inspiration

She later relocated to California to earn a master’s degree in mathematics at UC Berkeley, further refining her interest in intelligent systems.

Prior to ZeroEntropy, Houir Alami experimented with building a conversational AI assistant, an experience that highlighted the importance of providing accurate context to LLMs.

Advocating for Diversity in AI

Houir Alami, at 25, is a relatively rare female CEO in the deep infrastructure space for AI.

She encourages more women to enter the field, stating, “If you’re drawn to complex, technical problems, don’t let anyone make you feel like you’re not capable of pursuing them. You should go for it.”

Giving Back to the Community

She actively engages with her Moroccan roots by delivering presentations at high schools and universities, inspiring young women to pursue careers in STEM.