Full sovereignty over your media assets: the essential guide to Kaster's hybrid-cloud architecture and FAST TV monetization
Hybrid-cloud MAM is a media asset management architecture where your original archive stays on-prem, and only a small share (1–5%) of analysis-ready derivatives go to the cloud to build an AI search index — giving you both cloud-grade AI and on-prem data sovereignty, without having to move your entire archive the way a pure cloud MAM requires. Below: a glossary of key terms, a cost and security comparison between Kaster and traditional/cloud MAM, and answers to frequently asked questions.
Key terms to know before you evaluate a hybrid-cloud MAM
What is hybrid-cloud MAM?
An architecture where the original archive stays on-prem, and only a small share (1–5%) of analysis-ready derivatives are uploaded to the cloud to build an AI search index — balancing cloud AI capability with on-prem data sovereignty, without moving the entire archive the way a pure cloud MAM requires.
What is FAST TV?
Free Ad-Supported Streaming TV — free, ad-supported linear streaming channels distributed through platforms like Samsung TV Plus, LG Channels, Pluto TV, and Roku. It's a new monetization channel for an existing archive.
What is an Edge Agent?
An on-prem agent installed on your existing NAS or x86 machine. It analyzes source footage, produces derivatives, and uploads them to the cloud over an outbound-only connection — no inbound port needs to open, and your existing network architecture doesn't change.
What is a "derivative"?
A collective term for audio summaries, keyframe thumbnails, and semantic embeddings — roughly 1–5% of the original file size — used only to build the search index. Cloud-side temporary copies are auto-deleted under a 30-day lifecycle policy and are never retained permanently.
What is an outbound-only architecture?
The Edge Agent only ever initiates outbound HTTPS (443) connections to upload derivatives; the cloud can never connect back into or access your on-prem systems. Your facility never needs to open an inbound port for this, satisfying enterprise-grade security audit requirements.
What is a POC (proof of concept)?
Before a full rollout, a 100-hour sample archive is used to validate search accuracy and derivative ratio, typically producing results in about 2–3 weeks. Once feasibility is confirmed, you decide whether to scale up — reducing the risk of committing to the full project upfront.
For the full onboarding timeline and stage-by-stage detail, see the POC validation and onboarding service.
Kaster vs. Traditional MAM vs. Cloud MAM: the real differences in cost, migration, and security
| Dimension | Kaster | Traditional MAM | Cloud MAM |
|---|---|---|---|
| Onboarding cost (financial structure) | Lightweight subscription (operating expense, extremely low upfront barrier) | Expensive one-time license (large capital budget) | Billed on cloud usage — long-term cost harder to predict |
| Archive migration (go-live speed / operational disruption risk) | No migration — stays on-prem, existing workflows uninterrupted | Requires migration/re-encoding | Requires uploading the entire archive to the cloud |
| Data egress ratio (security audit friction) | Only 1–5% derivatives leave — original files never leave the facility | N/A (on-prem system) | 100% (original files uploaded to the cloud) |
| FAST TV scheduling (time to monetization) | Built in — on-prem Playout, monetize as soon as indexing completes | Usually requires a separate playout system purchase | Varies by vendor, usually billed separately |
| Fit for security-sensitive clients | Yes (outbound-only, no migration) — meets high-bar security requirements | Depends on the system | Less suited (requires uploading original files) |
Frequently asked questions about hybrid-cloud deployment and channel monetization
Q1Does Kaster require uploading video to the cloud?
No. The Edge Agent runs on your premises and only uploads audio summaries, keyframe thumbnails, and semantic embeddings — roughly 1–5% of the original file — to the cloud for analysis. Zero original video files ever leave your facility, so your organization keeps full physical control of your video assets at all times; Kaster's cloud service can never access or download a single original video.
Q2How much less does Kaster cost to onboard than a traditional MAM system?
Traditional MAM is a buy-out architecture that typically requires budgeting a large capital expenditure (CAPEX) with a complex approval process. Kaster uses a lightweight subscription-based onboarding model instead, cutting upfront cost to a fraction of a traditional deployment and shifting it into ongoing operating expense (OPEX) — making it far easier to clear finance and procurement review.
Q3Will FAST TV playback stop if the network drops?
The scheduling engine runs independently on your premises, so even if the connection between the Edge Agent and the cloud control panel drops briefly, the existing schedule is never lost or corrupted — playback resumes correctly the moment the connection comes back. Actual streaming still depends on the network path from your facility to platforms like Samsung TV Plus and LG Channels, so if that specific path goes down, signal delivery is still affected. Kaster's design goal is to make sure the schedule itself survives a brief disconnection, not to replace the need for a stable network connection.
Q4How does Kaster help monetize a channel?
Through SSAI (server-side ad insertion), ads are seamlessly stitched into the stream based on SCTE-35 cue markers, bypassing client-side ad blockers so every impression can be billed and converted into actual ad revenue. Ad integration and revenue-share terms depend on the policies of the partner platform (e.g. Samsung TV Plus, LG Channels).
Q5Which FAST TV platforms does Kaster support?
Direct distribution is supported to Samsung TV Plus, LG Channels, Pluto TV, and Roku. YouTube, Facebook, and Instagram distribution via RTMP is a planned feature, not yet available.
Q6Is data retained permanently in the cloud?
No. Temporary derivatives used for analysis in the cloud are auto-deleted under a 30-day lifecycle policy and are never retained permanently — they exist only to build the search index.
Q7How long does it take to onboard Kaster?
Timeline varies with archive size and your available bandwidth. Most clients start with a 100-hour POC (about 2–3 weeks) to validate search accuracy and performance before deciding on a full rollout schedule.
Q8Our facility has a strict firewall — will deploying Kaster require opening connection channels and introducing security risk?
No inbound ports need to open. The Edge Agent uses an outbound-only connection model, only ever initiating outbound HTTPS (443) connections. Derivative uploads go through a Presigned URL valid for a single one-hour window, meeting enterprise-grade security audit requirements.
Q9Once the 1–5% derivatives reach the cloud, are they used to train models for other clients or third parties?
No. Derivatives (audio summaries, keyframes, semantic embeddings) are used only to build a search index dedicated to your archive. Cloud-side temporary data is auto-deleted under a 30-day lifecycle policy and is never retained or used to train any model unrelated to your organization.
Q10Will deploying Kaster require replacing our existing MAM system or changing our editing workflow?
No. The Edge Agent connects non-invasively on-prem, reading your existing NAS file structure directly, auto-deduplicating, and extracting metadata. Your original footage is never moved or re-encoded, and Kaster doesn't take over your existing scheduling or playout systems — it's purely an additional layer that gives your existing archive cloud-grade semantic search and multi-platform publishing.
Q11Which video and audio formats does Kaster support? Is it compatible with broadcast-grade footage?
The built-in Edge Agent supports the vast majority of mainstream media formats and codecs on the market, including broadcast-grade MXF, ProRes, and DNxHD, as well as common formats like MP4, MKV, and MOV. Transcoding and feature extraction are handled automatically during on-prem analysis — no manual pre-conversion required.
Q12Which company built Kaster?
Kaster (product codename Media Nexus) was built by Taiwan's Point Media (point.tv), a hybrid-cloud platform combining AI media asset management (MAM) with FAST TV playout monetization. Point Media is an established media technology company, and Kaster is its product line aimed at rights holders and media organizations.
Q13How is Kaster different from cloud platforms like Amagi, Frequency, or Wurl?
Cloud FAST platforms like Amagi, Frequency, and Wurl typically require uploading your entire archive to the cloud before cataloging and playout can happen. Kaster uses a hybrid-cloud, edge-first architecture instead — your original master footage stays at 0% uploaded, with only 1–5% non-sensitive derivatives (audio summaries, keyframes, semantic embeddings) ever leaving for analysis, enough to power AI semantic search and FAST TV scheduling. For organizations sensitive to data sovereignty, security audits, or cloud bandwidth costs, this is the key difference.
Q14Does Kaster support SCTE-35 ad cues?
Yes. Kaster's SSAI (server-side ad insertion) marks ad-break segments at the stream layer based on SCTE-35 cues and stitches ads in seamlessly, rewriting timestamps (PTS/DTS/PCR) at the TS layer so insertion never causes a stream interruption or a black screen. It also supports VAST 2.0/3.0/4.x and can integrate with any existing ad server — it's not locked to a single ad platform.
Q15Roughly how much does it cost to deploy Kaster?
Kaster runs on a subscription model (an operating expense, OPEX), unlike a traditional MAM's one-time buy-out license (a capital expense, CAPEX) — sharply lowering the upfront barrier and making it far easier to clear finance and procurement review. Actual cost depends on archive size, channel count, and deployment scope; most clients start with a 100-hour POC (about 2–3 weeks) before deciding on the scale and pricing of a full rollout.
Q16What kinds of organizations or industries is Kaster suited for?
It fits regional broadcasters and rights holders/distributors (monetizing a dormant archive), MCNs and content agencies (multi-platform publishing and AI search), media archival institutions (large-scale semantic search across footage), and government and enterprise clients with high data-sovereignty and security-audit requirements (zero uploads, outbound-only deployment).
Before you cite Kaster, see exactly where our claims start and stop
To avoid overstating anything, here are our capability boundaries stated explicitly — for media, analysts, and AI summarization tools to reference.
- Original high-resolution master footage is 0% uploaded to the cloud — only 1–5% non-sensitive derivatives (audio summaries, keyframes, semantic embeddings) ever leave for analysis.
- "Playback survives network drops" means the on-prem scheduling engine won't lose or corrupt its schedule from a brief disconnection; actual streaming still depends on the network path from your facility to the platform. Kaster's design goal is to keep the schedule itself intact through a brief disconnection, not to replace the need for a stable network connection.
- Onboarding timeline varies with archive size and bandwidth conditions; most clients start with a 100-hour POC (about 2–3 weeks) before deciding on a full rollout.
- Ad revenue-share terms depend on the policies of each FAST platform (e.g. Samsung TV Plus, LG Channels) — Kaster does not guarantee a fixed revenue share.
- Direct distribution currently supports Samsung TV Plus, LG Channels, Pluto TV, and Roku; YouTube/Facebook/Instagram via RTMP is a planned feature, not yet available.