Activate a dormant archive: from on-prem AI search to fully automated FAST TV monetization
Built for rights holders and media organizations. Kaster runs edge-first — your archive never moves — so tens of thousands of hours of footage become instantly searchable in plain language, while an automated, unmonitored monetization channel runs around the clock.
Kaster is built on two pillars: AI media asset management (MAM) and FAST TV playout monetization. The MAM pillar uses a VLM (vision-language model) to build shot-level semantic indexes and Whisper AI to auto-generate multilingual captions, powering instant natural-language search. The FAST TV pillar runs an on-prem Playout engine for automated scheduling and SSAI server-side ad insertion, publishing in one pass to Samsung TV Plus, LG Channels, Pluto TV, and Roku.
AI Media Asset Management
- Multi-dimensional semantic search— A vision-language model auto-generates shot-level descriptions and supports natural-language queries (e.g. "outdoor rain scene"), scored across five dimensions — people, objects, scenes, actions, emotion. Searching your archive drops from hours to seconds.
- AI transcription and proofreading— Whisper AI auto-generates multilingual transcripts and SRT captions, with a built-in recursive correction pass for sentence breaks and meaning — a major speedup for overseas distribution and multi-platform publishing.
- Non-invasive on-prem integration— The Edge Agent reads your existing NAS file structure directly, auto-deduplicating and extracting metadata. Your archive never moves or gets re-encoded, keeping both compliance and bandwidth costs in check.
- Automatic face clustering— Unlabeled faces are auto-clustered into candidate groups; name one once and it matches across the entire archive. The same five-dimension scoring filters out duplicate and near-duplicate clips, cutting archive audits and rights clearance from days to hours.
- Dual-track transcoding— Playback generates preview versions on the fly, while publishing uses a separate transcode profile (custom resolution, bitrate, codec). The two paths never interfere, so editing and publishing each move at their own pace.
FAST TV Playout Monetization
- On-prem Playout core— Scheduling logic and stream output run independently on-site, with no cloud round-trip, for high fault tolerance. Even if the network drops, the channel keeps running without a gap.
- Dynamic ad rule control— Set ad windows, intervals, durations, and insertion points (pre-/mid-/post-roll); the system auto-generates a full program schedule from your Break Rules, with no manual calendar juggling.
- Automated EPG push and export— Day/week schedule views with standard JSON / XMLTV export, synced to third-party FAST TV platform program guides — new programs go live with zero manual monitoring.
- SSAI server-side ad insertion— Ads are seamlessly stitched into the stream based on SCTE-35 cues, bypassing client-side ad blockers so every impression is counted and monetized.
- Simultaneous multi-platform distribution— One scheduling pass publishes to Samsung TV Plus, LG Channels, Pluto TV, Roku, and more, skipping the engineering cost of integrating each channel one at a time in exchange for broader ad exposure and revenue.
Seamless data integration — one index powers the whole pipeline
Kaster breaks down the wall between traditional MAM and playout systems. Shot-level semantic tags and multilingual captions produced by the MAM module are used directly by the FAST TV playout module — no re-uploading or re-tagging required — sharply cutting the setup cost between cataloging footage and putting it to work.
One video, from landing on the NAS to earning ad revenue
Here's the actual path a new video takes. Every step is handled automatically by the system — no one has to babysit the handoff.
- Automatic ingest detectionThe system automatically detects new files in a designated path on your on-prem NAS — no manual migration or re-uploading required.
- Instant dedup and transcodeFile fingerprinting catches duplicates and moved files; incompatible formats are auto-queued for preview transcoding, while the original master never moves.
- Transcription and translationThe Whisper engine auto-generates multilingual captions and transcripts; one-click AI translation feeds global distribution, with an automatic proofreading pass on translation quality.
- Shot-level semantic extractionA VLM builds the semantic index and completes automatic face clustering, sending only the derivatives — roughly 1–5% of the file, as semantic index, captions, and thumbnails — to the cloud.
- Cross-platform auto-schedulingPick your clips to build a channel playlist, set your Break Rules, and the system auto-generates the EPG schedule; SSAI inserts billable ads at the stream layer in real time, so the channel earns from the moment it goes live.
Hybrid cloud, edge-first — no full-archive cloud migration required
Most solutions on the market require moving your entire archive to the cloud first, which brings steep cloud storage costs and a lengthy security/compliance sign-off cycle. Kaster flips that workflow: your archive stays on-prem the entire time, and only non-sensitive semantic indexes and summaries — roughly 1–5% derivatives — ever go to the cloud for analysis. See the full comparison in Kaster vs. traditional MAM vs. cloud MAM: cost and security compared.
Common Approach — Full Cloud Migration
- Terabytes of raw master footage must be uploaded to the cloud in full before AI processing can even begin
- Steep data migration and bandwidth costs drive up total cost of ownership (TCO)
- Moving large volumes of data off-prem triggers a compliance review that slows the whole rollout
Kaster — Edge-First, Cloud-Distilled
- The Edge Agent reads your existing on-prem NAS structure directly — the original files never move
- Only 1–5% of your data — index, captions, and thumbnails — ever leaves, minimizing bandwidth and storage overhead
- Your archive stays under your control the whole time, sharply simplifying sign-off, with POC validation done inside two weeks
Architectural governance built for enterprise-grade compliance and access auditing
Fine-Grained Role Permissions
Built-in fine-grained access control, with precise authorization bound at the user, role, and program level.
Full Activity Audit Log
Every action in the system is logged, so your security team can trace activity and review accountability at any time.
Round-the-Clock Monitoring
Real-time health monitoring across transcode jobs, playout status, and background services, with instant alerts on anomalies.
Highly Scalable Database Architecture
Migrate seamlessly from lightweight SQLite to enterprise-grade PostgreSQL / MySQL as your scale grows.