What if Atlas could handle the tedious parts of research while you focus on the thinking that matters?
Until now, Atlas was reactive. You'd ask a question, get an answer, then figure out your next step. That changes today. Agent mode lets you describe what you want done, then watch Atlas work through it autonomously.
Agent Mode
Toggle agent mode in the chat input and give Atlas a multi-step task. Something like: "Find all my notes on machine learning, identify common themes, and summarize them in a new note." Or: "Create a map connecting my research papers on climate change."
Then watch. Atlas searches your knowledge base, reads through relevant content, creates new notes, edits existing files, organizes your workspace. Every action is visible. You can follow along in real-time, and step in whenever you want to redirect or refine.
This transforms Atlas from a Q&A tool into a research assistant. The kind that can:
- Synthesize information across dozens of documents without you opening each one
- Draft structured notes from scattered research
- Build maps that visualize connections you haven't noticed yet
- Reorganize content based on themes that emerge from your work
You stay in control. You can pause, adjust, or take over at any point. But for the routine work (the gathering, sorting, connecting) Atlas handles it while you think about what it all means.
Guided Onboarding
New to Atlas? The first few minutes of any tool determine whether you'll stick with it or bounce. We've redesigned onboarding to make sure those minutes count.
New users now start with a structured tutorial that walks through the core workflow: uploading sources, creating notes, asking questions with citations, building maps. Instead of staring at a blank workspace, you explore with sample content already loaded. Real examples that demonstrate what's possible.
By the time you finish, you'll understand how Atlas thinks about knowledge: sources feed into chats, chats generate insights, insights become notes, notes connect into maps. The pieces fit together in ways that compound over time.
No more blank slate confusion. No more "what do I even do here?" You'll know exactly how to start building your knowledge workspace.
Improved
Several refinements make Atlas feel more responsive this week:
-
Faster suggestions and maps. We optimized how Atlas generates responses. Suggestions appear quicker. Maps build faster. The whole experience feels more immediate.
-
Bidirectional linking. Links between notes now stay in sync automatically. Rename "Research Notes" to "ML Research" and every note that links to it updates without you touching anything. Your connections stay intact as your thinking evolves.
-
Mobile past chats. On smaller screens, you can now browse your conversation history. Scroll through past chats, tap to continue any thread, pick up research wherever you left off.
-
Stop button. Works reliably when Atlas is mid-thought. Sometimes you realize you asked the wrong question. Now you can interrupt cleanly.
-
Chat @mentions. Mentioning notes in chat works consistently. Type
@to pull any note, source, or map into your conversation as context. -
Folder uploads. All files in uploaded folders now process correctly. Drag a folder of PDFs and they all land in your workspace.
-
Large document handling. Timeout issues on big files are resolved. Upload that 200-page PDF without worry.
-
Project selector. Displays properly on narrow screens. No more cut-off project names on smaller devices.
-
Send button. Responds immediately in chat. No more wondering if your message went through.
Have feedback? We'd love to hear from you at team@atlasworkspace.ai