A digital workshop for scholars, bringing research materials, emerging writing, and computational methods into a rigorous drafting environment.
Use Acquisitions to gather the materials you want to work with in Mira. You can upload files from your computer, scan a local folder for documents, or search open catalogues — currently the Internet Archive and Open Library — for works to import. Mira works with PDFs for now, and work is underway to support a wider range of formats.
Once materials are made available to Mira, their metadata must be verified and enriched. Academic articles must be carefully attributed to the correct author and publisher to avoid citational confusion when working with computational tools. While Mira does its best to do this automatically, users will need to manually verify and modify materials on a case-by-case basis to ensure their veracity.
The Main Library is where your processed documents live, all in one place. You can organise them with tags and browse them by Title, Author, Date, Publisher, and Source, with full detail available on every file. Each file becomes searchable once Mira has finished processing it, and this single library is what the Reading Room and the Atelier search against.
The Reading Room is where you both read and search your collection. Open any text to read it in full, or run a single search across everything in your library at once.
It is also where you annotate. A marginalia function lets you mark up your texts and make those notes available to Mira’s computational tools. Annotations are saved alongside your library, so they persist across sessions, and they are given to the language models as context but are never quoted as source text.
The Atelier is a research workspace and the heart of the application. Rather than a typical chatbot interface which provides one or two models and allows the user to upload a handful of documents, Mira’s Atelier provides a GUI-modifiable, model-agnostic, and multi-agent architecture for managing and coordinating large-language models with access to whatever materials you assign.
Every claim made by a model is backed by traceable excerpts from the Main Library. You can see what the model was given, what it found, and how it arrived at its conclusions before verifying the model’s claims against the original text; all without ever leaving the application.
A space for composition. Passages from reading and research can be moved directly into drafts. Excerpts carry their citations with them. When a passage enters a draft, its source information follows — so the bibliography builds itself as you write.