Datascene / VAA1
Book a 15-minute call

Research-led multimodal intelligence

Turn audiovisual archives into searchable, traceable knowledge.

Datascene helps archives and research institutions understand what is inside interviews, broadcasts, hearings, recordings, and audiovisual collections - while keeping every interpretation linked back to source evidence.

Honest status: Datascene is an active research software project. We are looking for a small number of serious early collaborators with real audiovisual material and a real analysis problem.

The problem

Many organisations hold thousands of hours of audiovisual material. The files may have titles, dates, and basic metadata, but the actual content remains hard to search, compare, interpret, and reuse.

Researchers still jump between videos, notes, transcripts, screenshots, and spreadsheets.
Archives often know what a file is called, but not what happens inside it scene by scene.
AI summaries can be useful, but they often detach claims from the original evidence.

The promise

Datascene turns audiovisual material into structured research objects: searchable, annotated, comparable, and linked back to transcript passages, timestamps, metadata, visual evidence, and analyst review.

How it works

Bring the material

Video, audio, transcripts, subtitles, metadata, field material, or archive samples.

Structure the evidence

Datascene links scenes, transcripts, speech, objects, annotations, metadata, and meaning structures.

Trace the claim

Move from an analytical observation back to the exact source moment, transcript segment, and supporting evidence.

Who it is for

Archives and memory institutions

For collections that need scene-level discovery, transcript-linked metadata, provenance-aware annotations, and better reuse.

Universities and research groups

For researchers who need transparent multimodal datasets, linguistic analysis, SFL-oriented schemas, and reproducible workflows.

Public-sector knowledge organisations

For hearings, consultations, policy communication, institutional memory, and multilingual governance-language analysis.

Creative and cultural organisations

For documentary archives, production memory, narrative alternatives, recurring themes, characters, and performance material.

Not for

Datascene is not a generic video summariser, stock media manager, or instant self-service AI app. It is for organisations with serious audiovisual material and a need for traceable analysis.

The Datascene USP

Interpretation you can trace, challenge, and reuse.

Generic AI tools stop at a summary. Datascene is designed for institutions that need to know why an interpretation was made, where the supporting evidence sits, and how that evidence connects across a collection.

Core advantage

Datascene turns audiovisual material into a navigable evidence graph, not a detached AI answer.

Every claim keeps its source

Interpretations stay connected to timestamps, transcript passages, visual cues, audio context, metadata, and analyst notes.

Collections become research structures

Material is organised scene by scene so teams can search, compare, annotate, and revisit the evidence behind a finding.

AI supports interpretation, not replacement

Datascene is built for accountable analysis where researchers can inspect how meaning was produced and challenge it when needed.

Current systems
Datascene direction
Black-box summaries
Claims linked back to exact source moments
Detached AI outputs
Interpretations with visible evidence trails
Flat file-level metadata
Scene-level structures across people, places, themes, and events
Separate transcript, video, audio, and spreadsheet workflows
One navigable layer connecting transcript, image, sound, metadata, and review

Founder note

Datascene grew from research needs in computational social science, digital humanities, institutional memory, and audiovisual analysis. The central question is simple: when machines begin to interpret culture, can we still understand where the interpretation came from?

Have audiovisual material that is difficult to analyse?

Tell us what kind of collection, archive, or research material you are working with. We are looking for a small number of serious early collaborators.

Prefer email? Contact: petteri.laine@helsinki.fi

FAQ

Is Datascene already built?

Datascene is under active development. The current goal is to work with selected collaborators on real audiovisual analysis problems, not to offer a generic public SaaS dashboard.

Is this free?

Initial conversations are free. Serious research consultation, tailored analysis, and implementation work are scoped separately based on the material, workflow, and organisational need.

Do we need to upload sensitive material?

Not initially. Early discussions can begin with descriptions, sample material, or non-sensitive examples. Datascene is being designed with local and controlled workflows in mind for restricted collections.

Who should contact Datascene?

Archives, universities, research groups, public-sector organisations, media teams, and creative organisations with audiovisual material that is difficult to search, analyse, compare, or reuse.