We’ve published “Augmented Research”, a guide meant to open up what Claude (or other LLM with the right harness) could mean for historical and social-science research once you move it past a chat window and let it work over your actual files. Starting as a three-hour internal workshop for non-programmers on the DISSINET team, it is growing into an open resource with over 100 documents. Written in close collaboration with AI, it reflects both human and machine contributions.
The workshop’s premise was simple: talking to an AI in a chat window and working with it over an actual folder of files are two different things. For document-heavy historical work: inquisition registers, charters, narrative sources and research layer notes, the second is often far more useful: it can read a whole folder of research documents, sources and data at once, keep project context across sessions, and manage research project over a corpus of observations and analytical code at the same time.
What began as a single 2.5-hour presentation – meant to convince researchers with a low-to-no programming background that AI harnesses built mostly to manage large code repositories, like Claude Code, can be very useful even to non-programmers – grew into something bigger: how-to material on Claude Code and Desktop, reflective pieces on how AI changes each stage of the research process, condensed cheatsheets for people who just want the short version, tutorials that take you from installing the thing to still using it a month later, and a section on institutional and regulatory context – Masaryk and Charles Universities guidance, journal and funder disclosure policies.
The core argument is a shift from prompt engineering to context engineering: building a persistent foundation: a project file, a folder Claude can read directly, a running record of decisions instead of phrasing the perfect question each time. That’s also what separates the two modes underneath. The model is the same either way; what differs is the harness around it. A chat window rebuilds context from scratch every time you open it. Working over a folder lets Claude hold your whole project, and even write the actual script behind an analysis. This can turn a spreadsheet into a reproducible interactive map or visualisation and make it a persistent part of a virtual research table.
OP JAK projekt „Za hranice bezpečnosti: role konfliktu v posilování odolnosti“ nabízí 3–8měsíční placené výzkumné pobyty zaměřené na DH analýzu středověkých latinských textů, konkrétně na systematický sběr a anotaci antiheretických nenávistných projevů.
Our new article published in Digital Scholarship in the Humanities uses the trial record of Bernard-Oth of Niort and his family to demonstrate the benefits of Computer-Assisted Semantic Text Modelling (CASTEMO), a custom approach that not only captures informational content but also its discursive and contextual embedding.
