Atlas Workspace
AI research assistant for understanding the literature
About Atlas Workspace
Atlas is a visual AI research workspace built around one idea: an answer about the literature is only useful if you can check it. You start with a research question, search scholarly literature for relevant papers, add PDFs, web articles, YouTube videos and notes to a project, and then ask questions across that material. Every answer carries citations that point back to the passage in the source that supports it, so you can confirm whether the evidence actually says what the answer claims instead of treating the response as a black box. The problem it targets is familiar to anyone writing a thesis or a literature review: chat tools are good at one-off conversations but lose track of sources, and note apps keep sources but leave all the reading and connecting to you.
What sets Atlas apart is its visual layer. Each source can be turned into an editable Knowledge Map that lays out its argument structure, cause-and-effect chains, processes, relationships and timelines with multi-level zoom, while a project-wide Semantic Map clusters your sources, notes and chats by meaning so you can see themes and connections across a body of work. Atlas positions itself as a persistent research workspace that combines a source library, visual maps and citation-level verification, rather than a chat window. It is built by AgentScale AI in Singapore, and the vendor states that research data is used to run the workspace and not to train AI models. Pricing is a simple two-plan ladder: a Free plan with up to 10 sources and 5 AI messages, and a Pro plan at $20 per month ($16 per month when billed yearly) with unlimited AI chats and sources.
You begin by creating a project around a research question. From there you can search a scholarly database the vendor describes as 200M+ research papers (with sources such as arXiv, PubMed / PMC, OpenAlex, Semantic Scholar and DOI / CrossRef), inspect the results, and add the relevant papers to the project. You can also bring in your own PDFs, web articles, YouTube videos (from their available transcripts), slides, documents and notes.
When a source is added, Atlas processes it into the workspace: it makes the content searchable, summarizes and structures it, indexes its citations, and builds visual maps. You can open a paper and generate a Knowledge Map to see how its claims, premises and evidence fit together, or ask about its methods, findings and unfamiliar concepts in chat.
Questions can be asked across the whole project, not just a single document. Atlas answers from the sources you have added, uses multi-step reasoning across papers, and shows the citation trail behind each claim so you can jump from an answer or a map node back to the exact supporting passage.
As your reading builds up, you compare findings across papers on the Semantic Map, turn highlights and ideas into notes that stay attached to their evidence, and develop cited synthesis notes for your literature review. When you need the work elsewhere, you export notes and download knowledge maps as images, and continue writing and managing references in your existing tools.
- •Paper search and discovery - Search scholarly literature directly from your research question, inspect the results and add relevant papers to your project, so your question and sources stay together. The Pro plan adds open-access paper discovery, and integrations cover arXiv, PubMed / PMC, OpenAlex, Semantic Scholar and DOI / CrossRef. Alongside papers, a project can hold PDFs, websites, notes, slides, documents and YouTube videos, so everything relevant to one question lives in one library.
- •Chat with grounded citations - Ask questions about a single PDF or across a whole project. Answers use multi-step reasoning over your papers, and each claim is explained with its source. Citations are auto-indexed on import, with a breakdown of how each one justifies the claim, and you can read a cited paper in context to see what it contributes to the argument. The Pro plan adds agentic research workflows and literature-indexed citations on top of the citation tracing available on every plan.
- •Knowledge Maps - Deconstruct a single source into an editable map that reveals its argument structure, causality, pros and cons, process flows, relationships and timelines. Multi-level zoom lets you focus on one concept, and every concept is tied back to the source text. You can navigate the map to follow how each claim builds up, which makes it practical for working through the reasoning of a long methods-heavy paper rather than skimming its abstract.
- •Semantic Maps - Visualize a whole project by meaning. Related ideas are grouped into clusters, the map can be projected from different angles and perspectives, and you can zoom into any paper to see how it connects to the rest of your literature. The map spans the sources, notes and chats in a project, so it reflects your own thinking as well as the papers, and on the Free plan Semantic Maps are available for all projects.
- •Summarizers for papers, articles and videos - Get the main ideas from a PDF, report, web article or YouTube lecture, then ask follow-up questions about methods, findings or what was said. A concept map maker generates an editable map from any single source.
- •Linked notes and synthesis - Turn highlights and ideas into notes that stay attached to their evidence, compare findings across sources, and build cited synthesis notes as your literature review takes shape. Notes and sources can be exported, and knowledge maps downloaded as images.
Atlas is aimed at students, graduate researchers, PhD candidates, labs and evidence-heavy professionals who need to read difficult papers, build literature reviews and keep track of what their sources actually say. A PhD candidate writing a literature review chapter can collect papers into one project, map the argument of each, and check every synthesized claim against the passage that supports it before it goes into the draft. Undergraduates working through dense readings can use the summarizers and chat to get the gist of a paper quickly and then ask about the parts they do not understand.
Research teams and labs that keep a shared reading list get a single place where sources, notes, questions and maps stay linked rather than scattered across PDFs and chat logs. Analysts and other professionals who need to get up to speed on a domain from fragmented documents can ask questions across a set of sources and follow the citation trail rather than trusting an unsourced summary.
It is also a fit for anyone who learns visually: the Knowledge and Semantic Maps show how a paper is built and how a field connects, which a linear chat transcript cannot. The Free plan is enough to try the maps and citations on a handful of sources, while ongoing, intensive research with many sources and unlimited AI chats needs Pro. Researchers handling sensitive material may value the stated policy that workspace data is not used to train AI models.
