Ought

What Ought does

Ought is a 501(c)(3) non-profit research lab, registered in Delaware and operating from San Francisco with a remote-friendly 8-person team distributed across the Bay Area, Austin, New York, and Oristà. Its stated mission is to develop mechanisms for delegating high-quality reasoning to advanced machine learning systems, grounded in a "Factored Cognition" / process-based ML architecture that decomposes complex reasoning into small, independently supervised sub-tasks (search, classification, summarization, extraction, critique). The organization's primary commercial product was Elicit, an AI research assistant that uses GPT-3 and fine-tuned T5 models to help academic researchers find, summarize, classify, and critique papers drawn from the Semantic Scholar API and Unpaywall; in September 2023 Elicit was spun off as an independent public benefit corporation that subsequently raised $9M in venture funding.

Ought's current business model is donation- and grant-based as a charitable non-profit (Tax ID 81-2121304), accepting contributions via credit card, PayPal, bank transfer, checks, wire, and cryptocurrency through Coinbase Commerce. Revenue is concentrated among three named grant funders: Open Philanthropy, Jaan Tallinn, and the Future of Life Institute, supplemented by individual donors in the effective altruism and longtermism communities. Since the Elicit spinoff, Ought's go-to-market is effectively community-led and research-oriented, distributing open-source tooling — the Interactive Composition Explorer (ICE) Python library and the Factored Cognition Primer tutorial — via GitHub, blog, newsletter, and Slack to build mindshare among AI safety researchers and practitioners rather than to monetize a product.

The customer base is concentrated in academic researchers (biomedicine, social sciences, economics) who use Elicit for literature review, with a secondary persona of non-expert users who lack access to expert-level research analysis. As of April 2022, Elicit had approximately 1,500 monthly active users and a 60% returning-user rate; that product and user base now sit with the spun-off Elicit PBC rather than with Ought itself.

Name

Ought

Legal name

Ought

Website

https://ought.org

Company type

Private

Founded year

2016

Operating status

Operating

Headcount range

1–10 employees

Short description

Ought is a 501(c)(3) non-profit AI research lab developing Factored Cognition and process-based ML tooling; it ships open-source libraries (ICE, Factored Cognition Primer) for AI safety researchers, funded by grants from Open Philanthropy, Jaan Tallinn, and the Future of Life Institute.

Ownership category

akta.pro rank

Product category

AI Research Assistant Software

NAICS

Scientific Research and Development Services (5417), Research and Development in the Social Sciences and Humanities (54172), Educational Support Services (611710)

SIC

Services-Prepackaged Software (7372), Services-Computer Programming Services (7371)

akta.pro primary industry

Automated Scoring, Grading & Feedback (AI/ML) Systems (EDAFADAE)

akta.pro secondary industries

Diagnostic & Adaptive Assessment Content (Skills/Readiness) (EDAGADAJ), Exam Prep & Standards-Aligned Review Courseware (Within Curriculum) (EDAFACAO)

Keywords

• AI research assistant
• Language model programs
• Factored cognition
• Literature review automation
• AI safety research

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Infrastructure, Marketing or Sales

Revenue model

1. Donations and Grants: Ought is a 501(c)(3) non-profit organization registered in Delaware. Revenue comes from charitable donations and grants from organizations aligned with AI safety and effective altruism. Tax ID: 81-2121304. Accepts credit card, PayPal, bank transfer, checks, wire transfers, and cryptocurrency (BTC, ETH, BCH, LTC, DAI, USDC).
2. Elicit (Spun-off Entity): Elicit has been spun off as an independent public benefit corporation and raised $9 million in venture funding (2023). This is now a separate entity from Ought.

Go-to-market motion2 records

Distribution channels2 records

Marketing channels5 records

Core offering

Ought is a non-profit research lab that develops mechanisms for delegating high-quality reasoning to advanced machine learning systems, primarily through Elicit—an AI research assistant that uses language models to automate literature reviews, including paper search, summarization, classification, extraction, and critique. It also builds the underlying open-source infrastructure (ICE library and Factored Cognition Primer) for compositional language model programs. Elicit has been spun off as an independent public benefit corporation as of September 2023.

Product overview

Ought operates as a product-driven non-profit research lab developing mechanisms for delegating high-quality reasoning to machine learning systems. The core product is Elicit, an AI research assistant for academic literature review. Supporting this are the Interactive Composition Explorer (ICE) Python library and the Factored Cognition Primer tutorial, which together enable the compositional language model programs underlying Elicit's architecture.

Differentiator

Problem solved

Functional benefit

Products and services

• Elicit AI research assistant that uses language models (GPT-3, T5) to automate parts of researchers' workflows, primarily for academic literature review. Helps researchers find papers, ask questions about them, summarize findings, classify study types, extract metadata, and surface methodological critiques. As of April 2022, Elicit had approximately 1,500 monthly active users with a 60% returning-user rate.
• Interactive Composition Explorer (ICE) Open-source Python library for writing, debugging, and visualizing compositional language model programs. Supports running recipes in different modes (humans, human+LM, LM), inspects execution traces in the browser, defines new LM agents, and parallelizes LM calls. Used by Ought as the underlying infrastructure for building Elicit.
• Factored Cognition Primer Interactive open-source tutorial that teaches how to write compositional language model programs. Covers amplification, debate, reasoning about long texts, decomposition, verification, tool use, and deduction using the Interactive Composition Explorer (ICE) library.

Quantifiable outcome

• 60% of monthly users are returning users, indicating strong product-market fit among researchers.
• +1 more outcomes

Named customers1 record

Segments2 records

Ideal customer profiles2 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration3 records

AI capability3 records

Feature5 records

Partnerships

Three partnerships are on record, tiered core and minor.

• Semantic ScholarcoreTechnology or IntegrationSemantic Scholar API is used by Elicit to retrieve paper titles, abstracts, citations, and metadata. Provides access to over 200 million abstracts and 66 million full-text papers.
• UnpaywallcoreTechnology or IntegrationUnpaywall is used by Elicit to retrieve PDF links for papers, enabling access to full-text content where available through open-access sources.
• CSET (Georgetown)minorStrategic or Co-development PartnerCSET researchers cited Elicit in three publications (Harnessed Lightning, Wisdom of the Crowd as Arbiter of Expert Disagreement, Classifying AI Systems). Ought worked closely with CSET as a research partner.

Scale indicators3 records

Recent moves7 records

Expansion highlights4 records

Direct peers

• Connected Papers: Connected Papers builds graph-based visualizations of academic papers related to a seed paper. It serves the same researcher audience and supports part of the literature exploration workflow that Elicit automates.
• Elicit: Elicit is the AI research assistant originally incubated inside Ought and spun off as an independent public benefit corporation in September 2023 with $9M in venture funding. It is the most direct comparison point because it is the same product, now operating as a standalone company.
• Scite: Scite applies AI to citation analysis, classifying whether citations support or contradict a paper's claims. It overlaps with Ought's critique-surfacing capability and serves the same academic researcher persona.
• Research Rabbit: Research Rabbit is an AI-driven literature discovery and mapping tool for academics. It targets the same literature-review pain point that Elicit/Ought addresses, with a different visualization-led approach.
• Consensus: Consensus is an AI-powered academic search engine that uses language models to surface answers from peer-reviewed research papers. It directly competes with Elicit/Ought on literature review and research-question workflows for academic users.

Emerging players

• Iris.ai: Iris.ai offers an AI engine for scientific research workflows, including literature mapping and analysis. It is an emerging player with partial overlap in target users (R&D teams, researchers) and use cases.

Broad incumbents

• Perplexity AI: Perplexity is a general-purpose AI answer engine with citation grounding. It increasingly competes with Elicit for researchers and knowledge workers who need AI-assisted, source-backed answers.
• Anthropic: Anthropic develops Claude, a general-purpose LLM with strong long-context reasoning capabilities. Claude competes directly with ChatGPT and emerging Elicit-style tools for AI-assisted research workflows.
• OpenAI: OpenAI provides the GPT-3 models that power Elicit's literature workflow and also ships ChatGPT, which overlaps with Elicit's research-assistant functionality. OpenAI is both a critical upstream dependency and a competitor.
• Semantic Scholar: Semantic Scholar (Allen Institute for AI) is a broad academic search engine that also serves as Ought's primary data partner via its API. It overlaps on paper discovery and is a foundational infrastructure provider in this space.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat3 records

Key risks6 records

Key highlights6 records

Customer concentration

• GitHubgithub.com/oughtinc
• Xx.com/oughtinc

Revenue estimate

Valuation estimate

Number of profiles

Profiles5 records

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 Ought