Essarion at a glance.

Essarion makes three products: ResearchAnything, Sourcipedia and Essarion Science. Each one is its own app, with its own website and its own sign-in. They grew out of the same research code, and two of them share a database. This page explains what each product does and exactly what they have in common.

01The products

All three do the reading for you and show their work. They differ in what you bring and what you get back.

ResearchAnything

A research assistant at researchanything.ai. You ask a question in plain words. It searches the web and academic databases, reads and compares the sources, and writes an answer with a citation on each claim. Around that sit projects for longer work, a library of saved sources, a citation tool and a developer API. Anyone can sign up, and every account starts on the free plan. Read more.

Sourcipedia

An encyclopedia at sourcipedia.com. Each entry is a plain-language summary written from sources, and the sources are listed with it. Entries are written and revised by a research pipeline that runs inside Sourcipedia. Reading and searching are free and need no account. Read more.

Essarion Science

A research scientist for biology and medicine at science.essarion.com. It plans a study, pulls records from public biomedical databases, runs the analysis in a live notebook, and tries to prove its own conclusions wrong before it reports them. It is a private deployment for now: there is no sign-up, and only accounts Essarion has set up can sign in. Read more.

Developer API

Essarion's developer API is part of ResearchAnything. You create a key in the ResearchAnything API Console and send questions to https://researchanything.ai/api/query. Keys belong to your ResearchAnything account, and API runs count against its plan. Read more.

02What they share

There is no central engine, gateway or account system that all three products call. Each runs as its own service. What they do share is smaller and more specific:

• Research code. All three are Python web services with React front ends, and each carries its own copy of the same research pipeline: plan the searches, fetch and read the pages, rank the sources, write with citations. Each copy is tuned for its product and deployed on its own.
• A database, between ResearchAnything and Sourcipedia. The two run against the same Postgres database. Sourcipedia writes the encyclopedia there, and ResearchAnything reads it as an index of already-vetted sources before it searches the web. The same database holds ResearchAnything's accounts, which Sourcipedia's administrators also sign in with.
• An optional shared sign-in. Essarion ID, shown as Sign in with Essarion, is one login you can use on products that offer it. Each product still keeps its own account record, linked to your Essarion ID.
• Legal terms. One Terms of Service and one Privacy Policy cover all of the products.

What is not shared

• Accounts. A ResearchAnything account does not give you access to Essarion Science. Sourcipedia has no reader accounts at all.
• Your work. Projects, runs, saved sources and papers stay in the product where you made them.
• Plans. The Free, Pro and Researcher plans belong to ResearchAnything and its API. Sourcipedia is free to read, and Essarion Science has no plans.

For what runs where, see Architecture.

03Models and data sources

The products use third-party large language models to plan, read and write. The Privacy Policy names the model providers as OpenAI and Anthropic. ResearchAnything gets its web results from Brave Search, and its academic results from scholarly indexes such as arXiv, PubMed, Europe PMC, Semantic Scholar, OpenAlex and CrossRef. Essarion Science also reads public biological databases. Section 4 of the Privacy Policy describes what each kind of provider receives.

Essarion's own models

Essarion also builds and trains its own models, outlined on the Research page: the BEM 1 family and Kinema, built for Essarion Science; DAI, in development for Sourcipedia; and small screening models, in development for ResearchAnything. The details are kept in-house for now, so these docs don't cover them.

04How to read these docs

The sidebar is grouped by product. Within each product, pages run in the order you are likely to need them: overview, then quickstart, then reference.

• If you're new, read Core concepts, then pick a product and follow its quickstart.
• If you're integrating, go to API quickstart and Streaming with SSE.
• If you're evaluating, read this page, Architecture, Security and Getting access.

05Where to go next

• Learn the terms these docs use: Core concepts
• See what runs where: Architecture
• Get into a product: Getting access
• Ask your first question: ResearchAnything quickstart
• Read an entry and follow its sources: Sourcipedia quickstart
• Call the API from your own code: Your first query