AI-powered natural language search in Preservica: transforming how your archival content is discovered
Helena Pilih
•
November 10, 2026
Helena Pilih, Head of Product at Preservica, explores the power of natural language search in archives and introduces Ask Your Archive, Preservica’s AI-powered search that is currently in preview with a small group of early access customers.
Keyword search only finds the exact match
Every archivist and records manager knows the gap. A researcher or colleague asks for records about previous head office locations, but the records only ever say “HQ”. Traditional keyword search matches words, not meaning, so searching on “head office” means that under-described and differently worded records are not found. I am sure you can think of many other examples.
In government, the consequences can be more critical: an FOI or public records request that misses relevant files is incomplete and therefore much harder to defend.
Natural language search closes that gap. It matches records by the meaning of your query, so you can search your archive the way you (or a colleague) would, using your own words to find what you are looking for.
Introducing Ask Your Archive
Ask Your Archive adds natural language search to Preservica’s standard search in our in New Generation editions. Once it’s switched on for your organization, choose “AI” mode and type a question in your own words.
It works from the text, titles and descriptions already in your records, across your archive, so there’s no extra cataloguing and it presents results in the standard search views, where you can filter them, customize the columns and download them as a CSV.
A relevance column shows how closely each result matches the meaning of your search query, and you can open every record to verify it. And because AI search returns actual files, not generated summaries, there are no invented phrases to second-guess.
The first version is currently in preview for a small number of early access customers that are helping to shape the feature ahead of general release early next year. And like our other embedded AI tools, you can switch Ask Your Archive on or off, so you stay in control.
Your permissions still apply to every result
AI search must never mean wider exposure. For every query, Ask Your Archive operates within your existing security tags, roles, and permissions, exactly as keyword search does. If a highly relevant record sits outside your access, it never appears in your results.
That makes Preservica’s natural language search safe for government, corporate, and other sensitive collections, where the wrong result in the wrong hands can be a compliance or privacy risk.
AI-powered natural language search example: government records request
Embedded AI with a lighter footprint
Ask Your Archive isn’t a bolt-on general purpose chatbot. It runs inside your Preservica environment, using a small open model built for search rather than a large generative one. Your records stay within Preservica. Meaning is extracted once, at ingest, and your records never leave the platform neither are they used to train any models.
This design keeps compute demands relatively low, in line with our sustainability commitments. It also removes the time-consuming export and re-import steps that external AI tools require, along with the risk they bring.
Trusted content makes trustworthy AI
AI is only as good as the content it reads. Ask Your Archive only searches your own repository. It doesn’t pull answers from the internet, and it returns your records, not generated text.
Preservica ensures your content becomes the trusted foundation for high-quality AI search results, holding authentic, well-described records, protected by Active Digital Preservation™ so they stay readable and their integrity is continuously checked over time. Your content stays governed, auditable, readable and under your control.
Human-centric AI that widens what you can find
Ask Your Archive searches the text extracted from your records, so the more of your archive that carries reliable text, the more it can find.
Our other embedded AI tools enable you to enrich your digitized, image and AV content, with you always in control:
AV transcription and OCR turn recordings and scanned pages into searchable text.
PII detection flags sensitive content, so you can protect it with security tags
Metadata quality control standardizes descriptions for consistent, dependable results.
Image description adds metadata to your photographs
In line with the EU AI Act, we label AI-generated metadata, such as image analysis, PII detection and transcription results, so you always know where it came from.
What’s next?
Ask Your Archive is currently in preview with a small number of early access customers that are helping to shape and refine the solution ahead of release early next year.
Learn more
