What we're building

Imagine you're researching everyday life in 18th-century Ottoman trade cities. The sources you need might be scattered across dozens of archives, written in Ottoman Turkish, Greek, Arabic, Hebrew, French, and Italian. Finding them means searching each archive individually, often in a language you don't speak, hoping your search terms catch what's there.

Archivum is a research tool that does this work for you. You describe your research question, and it searches public archives in parallel — across languages, collections, and countries. It returns a list of relevant primary sources, each with a summary, an explanation of why it's relevant, and a direct link to the original.

It's a research assistant for source discovery, not a replacement for your expertise. You still read, interpret, and write. Archivum just makes sure you don't miss what's out there.

education time history learn study book countries world

How you can help

Take our survey (~10 minutes)

Tell us about your research workflow, the archives you use, and the challenges you face when searching for primary sources. Your responses will shape the tool from the ground up.

Join the waitlist

Be the first to know when Archivum is ready to try. We'll reach out when we have something to show.

Who we are

We're researchers from the Tübingen AI Center working on AI for information seeking. We believe that the most interesting research happens when you build tools for real problems, not models nobody uses and benchmarks that are outdated the minute they are published.

We chose to work with historians because your research demands exactly the kind of cross-lingual, cross-archive discovery that AI should be good at but isn't yet.

This is an open research project. The tools and models we build will be openly available.

Why we need historians

We're a research team at the intersection of machine learning and the humanities. We could build this tool based on our own assumptions, but we'd almost certainly get it wrong.

We want to understand how you actually work:

Which archives do you search, and how?
What languages do you encounter that you can't read yourself?
Where do you spend the most time on tedious discovery work?
What would actually help — and what would get in the way?

Your answers will directly shape what we build. This isn't a finished product looking for users. It's a research project looking for collaborators.

business start up company goals marketing strategy plan

FAQ

Frequently asked questions answered

Is this an AI that writes history?

No. Archivum helps you find primary sources. It doesn't interpret them, write narratives, or draw conclusions. That's your job. We just want to make the search easier.

What archives does it search?

We're starting with large public collections like Europeana and WorldCat, and expanding based on what historians actually use. That's one of the things the survey asks about.

What languages does it support?

The tool will work across many languages, but we want to understand which languages matter most to you so we can prioritize and test thoroughly.

Do I need to be a professional historian?

No. If you work with historical primary sources, whether as an academic, a student, or an independent researcher, we want to hear from you.

Is this free?

The research and the models we develop will be open source. We're building this as a research project, not a commercial product.

How is my data used?

Survey responses help us understand historian workflows and design the tool. We won't share your individual responses. Waitlist information is only used to contact you about Archivum.