Self-hosted media intelligence
A table of contents for everything you shoot.
Mokuji indexes every part of your footage, so you find the exact moment in seconds instead of hours.
Your best shot is in there. Somewhere.
Years of footage. Filenames like
A001_C047.mov. The moment you need is buried in hour
forty-seven of dailies, and the only index is whoever happened to log
it, if anybody did.
Scrubbing and guesswork.
Hours of manual logging, tribal knowledge, and re-watching clips to find one line of dialogue or one face. The archive is technically searchable and practically not.
“Chef plating food” → the clip.
Describe the moment in plain language. Get back the in and out points, ranked across everything Mokuji has read.
Many engines. One index at the end.
Each engine reads a different signal out of the same frames, and more are added as Mokuji grows. Nothing is searched in isolation: every query is scored across all of them together, which is how it answers searches a filename never could.
Describe reads what is on screen. Speech reads what is said, and by whom. Identity recognizes the people you enroll. Detect finds objects, your logos and on-screen text. Index fuses all of it into one place you can search.
Searches a filename could never answer.
Every one of these is a single query, typed the way you would ask a colleague.
“Every time the CEO mentions the merger”
Speech Engine finds every line that says it, and Identity Engine knows who the CEO is from the profiles you keep, so you can search by role instead of by name. Returns the lines, not the files that contain them.
“Maria’s daughter talks about her first job”
Identity Engine knows how the people you enroll are related, and Speech Engine finds where she talks about it. Nobody has to remember her name.
“Close-up of Maria Alvarez”
Identity Engine finds every shot she is recognized in, and Describe Engine tells a close-up from a wide. Each result lists who else is on camera with her.
“Every time the Acme logo is on screen”
Detect Engine matches the logos you enroll from a few reference images, on title cards, end slates and graphics, and returns each appearance with its timecode.
“Campos de lavanda”
Describe Engine searches by what is in the frame, not by keywords, so a search in Spanish finds the same lavender fields a search in English does. No tags in two languages.
“Q3 roadmap”
Detect Engine reads the printed text on screen, from slides and signs to lower thirds and credits, so those words are searchable even when nobody logged them.
From raw footage to searchable, in four steps.
Drop media in. Mokuji handles the rest, automatically, on ingest. Nobody has to remember to run anything.
Ingest
Watch a folder or upload. Proxies are generated into a folder you choose; originals are indexed in place on your storage or copied into a managed store.
Enrich
Every engine runs across the asset, each reading its own signal from the same frames.
Index
Every signal is fused into one searchable index, cross-referenced by identity and by time.
Find and cut
Search in plain language, review and rate selects, then export straight into your edit.
Secure by design.
Mokuji runs on your own infrastructure, so your footage and the face and voice data derived from it stay under your control and inside your existing security boundary.
The security page covers the detail, including the limits, so your team can check it rather than take our word for it.
At a glance
- Deployment
- Inside your network, on hardware you own; cloud processing is optional and off by default
- Telemetry
- No telemetry, no analytics and no license check-in
- Air gap
- Built to run with no internet connection once installed; models are staged locally
- Biometric data
- Stays on your storage, under your retention
- Access
- Role-based permissions; collections shared by team or user
- Media URLs
- Signed and expiring, not open file paths
One index, many rooms.
The pain is the same everywhere: the moment you need is in there, and nobody can find it fast enough. What changes is what the moment is worth, and who is allowed to see it.
Post-production
Dailies, selects and archive in one searchable index.
Broadcast & sport
Turn live and post archives into instant highlights.
Enterprise knowledge
Training, meetings and all-hands, searchable down to the moment.
Corporate archives
Decades of brand footage, finally usable.
Government & defense
Evidence and mission media that must stay under your control.
Legal & investigations
Interview libraries searched in-house, with timecode.
Worship & nonprofits
Years of services and testimonies, made searchable.
Education & research
Lecture capture and field recordings turned into a knowledge base.
Index your library.
Bring a slice of your own footage and we will index it in front of you. If it does not fit your archive, we would rather tell you early.