FirstPrinciples

What FirstPrinciples does

FirstPrinciples is an independent research company based in Toronto, Canada, building AI systems exclusively for discovery in fundamental science. It targets the global scientific research community — individual physicists, mathematicians, and research institutions — that is constrained by what the company describes as a 'growing burden of knowledge' in which human cognition cannot scale with the proliferation of available information. The company's stated mission is to advance AI-driven discovery in fundamental science with knowledge treated as a public good, distinguishing itself from both academic human-led discovery workflows and commercial AI labs with proprietary priorities.

The company's core offering is Theo, a modular 'AI Physicist' system comprising 120B+ parameter specialized fine-tuned models across physics domains, trained on a curated corpus of 3M+ high-quality scientific papers. The product portfolio spans three coordinated products: the Theo AI Physicist infrastructure for reasoning like a physicist; Theo Collaborator, a domain-specialized scientific workflow system for planning, executing, and revising research; and Theo Conjecture, an automated conjecturing system for mathematical structure currently in Research Preview. Key technical capabilities include hypothesis generation, symbolic reasoning, tool integration, validation loops, Dynamic Research Objects (DROs) for reproducible research steps, and CiteVerify for fake citation detection. The AI Physicist reached a first autonomy milestone ahead of schedule in 2025, enabling self-guidance from research question to hypothesis to evaluation, with internal experiments running on quantum information questions.

FirstPrinciples currently operates as a research-stage organization with no publicly disclosed pricing, no customer logos, and no announced revenue. Its go-to-market is product-led growth via self-serve early-access signup for Theo Collaborator and Theo Conjecture through the company website, paired with content marketing through its blog ('Hub') and participation in academic conferences such as the NeurIPS Machine Learning and the Physical Sciences (ML4PS) workshop. Distribution is global via digital channels. The company is listed as a non-profit, has 11-50 employees, and discloses no funding rounds, investor base, or ownership structure in the available data.

Name

FirstPrinciples

Legal name

FirstPrinciples

Website

https://firstprinciples.org

Company type

Private

Founded year

2024

Operating status

Operating

Headcount range

11–50 employees

Short description

FirstPrinciples is an independent Toronto-based research company building AI systems (Theo AI Physicist, Theo Collaborator, Theo Conjecture) for discovery in fundamental science, serving the global scientific research community through early-access products.

Ownership category

akta.pro rank

Product category

AI for Scientific Research

NAICS

Software Publishers (5132), Computer Systems Design and Related Services (5415)

SIC

Services-Prepackaged Software (7372)

akta.pro primary industry

AI Application Enablement Platforms (Copilot/Agent Frameworks, SDKs) (HDAEANAJ)

akta.pro secondary industry

Enterprise AI Data & Knowledge Platforms (Vector Databases, Knowledge Graphs) (HDAEANAH)

Keywords

• AI for science
• Scientific research AI
• Automated hypothesis generation
• Mathematical conjecturing tools
• Scientific workflow automation

Headquarters

HQ city

Toronto

HQ country

Canada

HQ region

North America

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Operations

Revenue model

1. AI Research Platform Access: FirstPrinciples appears to be in research/pre-revenue phase, offering early access to Theo Collaborator and Theo Conjecture products. Revenue model not yet publicly disclosed as the company focuses on advancing AI-driven discovery in fundamental science.

Go-to-market motion1 record

Distribution channels1 record

Marketing channels4 records

Core offering

FirstPrinciples builds AI systems specifically for discovery in fundamental science, centered on its Theo product family. The portfolio includes Theo, a modular AI Physicist system built on 120B+ parameters across 5 specialized fine-tuned models trained on 3M+ scientific papers; Theo Collaborator, a domain-specialized scientific workflow system for planning, executing, and revising research; and Theo Conjecture, an automated conjecturing system for exploring mathematical structure.

Product overview

FirstPrinciples is a research company building AI systems exclusively for discovery in fundamental science. The product portfolio centers on Theo, an AI Physicist system that provides modular infrastructure for AI-enabled scientific discovery, complemented by Theo Collaborator (a domain-specialized scientific workflow system for planning and executing research) and Theo Conjecture (an automated conjecturing system for mathematical exploration). The three products are designed to work together, with Theo providing the core reasoning infrastructure while Theo Collaborator handles workflow management and Theo Conjecture enables automated mathematical conjecture formation.

Differentiator

Problem solved

Functional benefit

Brands

• Theo: The AI Physicist - A modular system built to reason like a physicist for AI-enabled discovery in fundamental science, with 5 fine-tuned specialized models and 120B+ parameters trained on 3M+ scientific papers.
• Theo Collaborator
• Theo Conjecture

Products and services

• Theo - AI Physicist A modular AI system built to reason like a physicist, serving as infrastructure for AI-enabled discovery in fundamental science. Features 5 specialized fine-tuned models with 120B+ parameters trained on a curated scientific corpus of 3M+ papers, with hypothesis generation, symbolic reasoning, tool integration, validation loops, and Dynamic Research Objects (DROs) for reproducible research.
• Theo Collaborator A domain-specialized scientific workflow system that enables researchers to plan, execute, and revise research without starting over. Available through early access signup for the scientific research community.
• Theo Conjecture An automated conjecturing system for exploring mathematical structure. Currently in Research Preview with early access signup available for researchers.

Quantifiable outcome

• First autonomy milestone achieved ahead of schedule for AI Physicist system
• +3 more outcomes

Segments1 record

Ideal customer profiles2 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability5 records

Feature5 records

Scale indicators4 records

Recent moves5 records

Expansion highlights4 records

Direct peers

• Sakana AI: AI research lab focused on building foundation models specifically for scientific discovery, including automated research and AI-driven science. Most directly comparable to FirstPrinciples in mission, product architecture, and customer base.
• Lila Sciences: AI-for-science organization building autonomous scientific research systems across multiple scientific domains. Comparable to FirstPrinciples in developing AI agents that conduct and accelerate scientific experimentation.
• FutureHouse: Non-profit research organization building AI agents (e.g., Crow, Falcon) to automate scientific literature review and discovery tasks. Closely mirrors FirstPrinciples' public-good, scientific-AI-agent focus and non-profit structure.

Broad incumbents

• DeepMind (Alphabet): Frontier AI research lab with significant work in AI for science, including AlphaFold and other scientific reasoning systems. A broad incumbent whose general-purpose models overlap with FirstPrinciples' scientific reasoning ambitions.
• Anthropic: Frontier AI lab whose Claude models are increasingly used by researchers for scientific reasoning and literature tasks. Broad incumbent overlapping with FirstPrinciples' scientific use cases through general-purpose models.
• OpenAI: Frontier AI lab offering general-purpose reasoning models (including o-series) used widely for scientific research and hypothesis generation. Broad incumbent that competes for the same scientific AI workloads with general-purpose models.
• IBM Research: Long-running industrial research organization with extensive AI-for-science efforts across physics, chemistry, and materials. Broad incumbent in AI-for-science with institutional research culture comparable to FirstPrinciples' positioning.
• Microsoft Research: Major industrial research lab with substantial AI-for-science initiatives including AI2, materials discovery, and scientific reasoning tooling. Broad incumbent with overlapping scientific AI ambitions.

Emerging players

• Periodic Labs: Early-stage company applying AI to scientific experimentation, including autonomous labs and scientific agents. Emerging player with partial overlap with FirstPrinciples' autonomous scientific discovery thesis.
• Citrine Informatics: AI platform applying machine learning to materials science and chemistry R&D for industrial customers. Emerging player in domain-specialized scientific AI, adjacent to FirstPrinciples' fundamental-science focus.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks6 records

Key highlights6 records

Customer concentration

• LinkedInlinkedin.com/company/firstprinciplesfoundation/?
• Xx.com/FirstPrincip1es
• Facebookfacebook.com/FirstPrinciplesFoundation

Revenue estimate

Valuation estimate

Number of profiles

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about FirstPrinciples