Unlocking Hidden Stories: How AI Is Opening Up Central Baltic Archives

Whether someone is tracing their family roots, researching local history or simply curious about the past, finding the right document in an archive can feel like searching for a needle in a haystack.

Across Finland, Estonia and Latvia, archives hold millions of historical records. But even as more material is being digitised, much of it remains difficult to search. Documents are handwritten, written in different languages and spread across vast collections. The Central Baltic project ArchXAI is working to change that by developing practical AI tools that make historical information easier to find and use.

Teaching AI to read the past

One of the project’s key goals is to help computers read old handwritten documents. The idea sounds simple, but the challenge is enormous: many records were written centuries ago in styles that even modern readers struggle to decipher.

To train the AI, experts manually transcribe historical texts word by word and character by character. The system then learns to recognise handwriting patterns and convert scanned images into searchable digital text.

The project is developing models for Finnish, Swedish, Estonian, Latvian and Russian materials, helping to open up records that were previously difficult to access. To ensure lasting impact, the AI models developed in the project will be openly published and documented, allowing other archives and organisations to benefit from the results long after the project has ended.

From endless searching to smarter discovery

Reading old documents is only part of the puzzle. ArchXAI is also creating AI-based tools that can automatically identify names, places and key details in archival records.

Today, archive professionals often receive requests from people looking for information about a relative without knowing which archive or collection might hold the relevant records. Finding the answer can require searching through multiple databases, catalogues and digitised documents, often across different languages. The process is time-consuming and relies heavily on the expertise of archive specialists.

The project’s AI-powered search and indexing tools could dramatically speed up this work. By enriching digitised records with automatically generated metadata, archive professionals will be able to locate relevant documents more quickly through a single interface, reducing the need for manual searches across multiple collections.

For ordinary users, the benefits are just as important. Someone researching their grandparents, for example, may no longer need to know exactly which archive holds the information before starting their search.

A shared history across borders

The cross-border cooperation between Finland, Estonia and Latvia is at the heart of the project. The three countries share centuries of intertwined history, from the Swedish Empire to the Russian Empire and, for Estonia and Latvia, the Soviet period.

That shared past means their archives often contain related information and similar types of records. By developing common solutions together, the partners can pool expertise, create larger training datasets and build tools that benefit archives across the whole region.

The most interesting part of the project so far has been the opportunity to discuss and exchange experiences with colleagues from the National Archives of Estonia and Latvia. During our visit to Tartu and Riga in February, we quickly found common ground and recognised many of the same challenges in our daily work. Meeting such knowledgeable and dedicated archive professionals also gave me an even greater sense of pride in our profession, says researcher Sanna Joska from the National Archives of Finland.

AI supporting people, not replacing them

Although artificial intelligence is central to ArchXAI, the project is built on the expertise of archive professionals. As project manager Anna Ollanketo from the South-Eastern Finland University of Applied Sciences (Xamk) notes, archives rely on deep professional knowledge and experience. The AI tools are designed to support that expertise, not replace it.

Coming from completely outside the archival field, one of the most fascinating aspects of the project for me has been getting to know the experts responsible for handling information requests in archives. They know their archives, databases and collections inside out, and are able to combine technical search skills with historical context and the interpretation of multilingual texts and old handwriting.

What has surprised me most is the strong professional identity and genuine enthusiasm these experts have for their work. Archival work is often seen as quiet work behind the scenes, but in reality, it involves demanding information retrieval and problem-solving, with every request presenting a unique challenge. What has particularly stayed with me is the passion and positivity with which these professionals talk about their work and its importance, continues project manager Anna Ollanketo from Xamk.

Opening archives to everyone

Looking ahead, the ArchXAI team sees a future where archives are easier to use for everyone, from researchers and students to family historians and curious citizens. By making historical records more searchable across languages and borders, the project is helping more people connect with the stories hidden in the region’s shared past.

More information

ArchXAI | Project webspace | Central Baltic Programme

Archives extended by AI | South-Eastern Finland University of Applied Sciences