Activeloop Scientific Discover: The AI Analyst for Scientific Research

Hey 👋 Davit here, founder of Activeloop.

Today we are introducing something we believe will fundamentally reshape how scientific discovery happens.

Activeloop’s Scientific Discover, an intelligence agent built on one of the largest datasets of indexed scientific research. Here are the details:

https://www.youtube.com/watch?v=x8Lv5-C9ntw

The Problem: Scientific Data Is Broken

Every scientist already knows the pain.

• Critical insights are buried across PDFs, images, tables, and supplementary files.
• Key figures live inside screenshots, not structured databases.
• Twenty years of research is trapped in incompatible formats and outdated systems.
• Every new project begins with weeks of manual data wrangling.

The White House recently launched the Genesis Mission, recognizing that fragmented scientific data is one of the biggest blockers to breakthroughs in drug discovery, materials science, climate modeling, and more.

The Genesis Mission aims to build an integrated AI platform that harnesses Federal scientific datasets. The goal is to train scientific foundation models and create AI agents that test hypotheses, automate research workflows, and accelerate scientific breakthroughs.

The country is trying to unify scientific data for AI.

We decided to show what that future looks like.

Introducing L1: The Scientific Data Agent

To demonstrate what becomes possible when scientific data is indexed natively for AI, we built Activeloop L1, a multimodal scientific research agent running on top of:

• 25M open access papers
• 450M+ plus pages
• 175TB of fully visual indexed scientific data

Unlike traditional NLP or paper search, L1 sees science.

It understands:

• charts
• molecules
• protein structures
• equations
• experimental tables
• clinical graphs
• and the text that surrounds them

This allows L1 to answer questions no text only model can touch.

The Results: A New State of the Art 48% SOTA on Humanity’s Last Exam with Tools

L1 (Gemini 3 Pro) outperforms top models including Grok Heavy with tools, and GPT 5 Pro with tools.

Multimodal scientific reasoning

L1 synthesizes biochemical structures with clinical findings to answer end-to end research questions.

Example Query

Which compounds show novel synergy with metformin for treating type 2 diabetes?

L1 scans molecular structures, clinical outcomes, and experimental tables at the same time and produces grounded, citation linked answers.

Why This Matters

Multimodal scientific agents unlock capabilities that were not possible before.

This extends far beyond drug discovery. It applies to materials science, climate research, algorithmic innovation, and more.

The applications are enormous.

Try L1 Today with Our OpenAI Compatible API

Get Started

Try the Science Agent: https://chat.activeloop.ai/science

Docs and API: https://docs.activeloop.ai

Build with the OpenAI compatible API today.