Understand
Transcripts, scenes and visual metadata become a searchable representation of the archive.
ARCHIVE INTELLIGENCE · AI-NATIVE MEDIA
Turn existing footage into the next story.
Reelume is building an AI-native media intelligence platform that turns large video archives into new digital content. It understands transcripts and visual metadata, retrieves relevant moments, and uses Claude as the reasoning layer for editorial intent, evidence-grounded story planning, selection decisions, and quality review before human approval.
Early-stage · Bootstrapped · Seoul
THE PROBLEM
The best moment in an archive is useless if nobody can find it.
Retrieve
Natural-language briefs surface relevant moments with source context and provenance.
Reason
Claude connects evidence to editorial intent, narrative structure and review criteria.
PRODUCT CONCEPT
From a question to an evidence-grounded story plan.
A working product direction — shown as concept UI, not a claim of released functionality.
Archive IntelligenceCONCEPT UI
How has the city changed over the last decade? ↵
142 relevant moments · ranked with source context
Riverfront regeneration
Public space, infrastructure, cityscape
Urban mobility
Transit, streets, changing movement
New city infrastructure
Present-day visual comparison
CLAUDE · STORY PLAN
- 01OpeningEstablish then vs. now
- 02TransformationConnect evidence across years
- 03PeopleGround change in lived context
- 04ReviewVerify sources before approval
WHY CLAUDE
Claude is the reasoning layer — not the archive.
Retrieval finds candidate evidence. Claude reasons over that evidence to interpret editorial intent, build an evidence-grounded story plan, explain selection decisions, and apply quality-review criteria before human approval.
Interpret audience, format, tone and editorial objective.
Connect retrieved moments without losing source context.
Explain selection, ordering and continuity choices.
Apply review criteria before a human approves the output.
BUILT FOR RESPONSIBLE MEDIA WORKFLOWS
Source provenance
Every recommendation should remain traceable to its originating media.
Authorized media
Designed for archives and media the operator is authorized to use.
Human approval
AI accelerates discovery and reasoning; publishing remains a human decision.
ABOUT REELUME
A new way to work with video archives.
Reelume is an early-stage, bootstrapped AI media technology project focused on making large video collections searchable, understandable and reusable. We are developing the product from Seoul, South Korea.
- Stage
- Early-stage / Bootstrapped
- Started
- 2026
- Location
- Seoul, South Korea
- Contact
- founder@jongsik.com
BUILDING NOW
Make archives useful again.
Interested in archive intelligence, AI-native editorial workflows, or early product collaboration?
founder@jongsik.com ↗