Its own workflow engine
ShippingTrigger → steps → approval → result runs, declared once and executed by the Node itself — not a script someone has to remember to kick off.
The gap
Your media operation shouldn't run on someone's laptop.
Most teams already have the pieces — storage, scripts, a review platform, automations someone built. What they don't have is a machine whose job is to keep all of it running.
01
The ingest automation lives on an editor's MacBook. They close the lid at 6pm and the pipeline stops until morning.
02
Processing eight terabytes means pushing eight terabytes to a cloud worker first. The egress bill and the wall clock both make it a non-starter.
03
The NAS holds nine years of footage. Nothing is continuously indexing it, transcribing it, or noticing what arrived last night.
04
There are scripts. They work, mostly. Nobody would call them a system, and nobody wants to be the person who owns them.
05
Every AI experiment ties up a workstation for hours, so it only runs when someone remembers to run it.
Media Node exists to turn that work into infrastructure: one machine, inside your environment, that is responsible for it and stays responsible for it.
Where it sits
Put the automation where the media already lives.
The Node sits on the same LAN as your storage. Work happens next to the files instead of across an internet connection.
Source
NAS / shared storage
SMB and NFS volumes, mounted read-only. Your media never moves.
Compute
Media Node
Watches the volumes, runs the queue, does the processing. Always on.
Intelligence
Media Nexus + workflows
Catalogue, search, policy, and approvals — running on the Node itself.
Destinations
Cloud / archive / team
Review platforms, archive tiers, and every editor who needs the shot.
Local by default
Media is read over your own network at your own line rate. Large originals do not leave the building to be processed, and derivatives are written to storage you designate.
Always on
The Node runs as a managed system service. It comes back after a reboot or a power loss, remounts storage, and resumes interrupted work without a keyboard attached.
Bounded on purpose
Source and archive locations are mounted so automation cannot write to them. Anything that modifies media stops for a named approval first.
Yours to control
There is no inbound listener and no remote shell. The Node reaches out on an authenticated connection and performs only the typed actions your policy allows.
The difference between a proxy pass that finishes overnight and one that never gets scheduled is usually not the software. It is whether anything was awake and connected to the storage when the footage landed.
What it does
Three jobs, running continuously.
Understand the media, move the media, keep the operation running. This is the processing layer — the automation platform underneath it is further down the page. Every line below is marked with where it actually stands — we don't ship roadmap as though it were product.
job · proxy
Proxy generation
Every clip that lands gets an edit-ready proxy written to your derivatives volume.
job · transcript
Transcription
Footage is transcribed on the Node, so interviews are searchable by what was said.
job · index
Indexing & metadata
Technical metadata, thumbnails, and catalogue records — kept current, not batch-run.
job · verify
Backup verification
Backup Guard confirms archive copies actually exist where policy says they should.
job · deliver
Uploads & delivery
Finished work moves to review and delivery destinations on a schedule, not a reminder.
Shipping in the current release
Early access running with pilot deployments
Planned on the roadmap, not yet delivered
Understand your media
- Continuous discovery of new and changed files Shipping
- Technical metadata extraction and normalization Shipping
- Proxy generation into declared derivative storage Shipping
- Local transcription of completed media jobs Shipping
- Transcript search that runs entirely on the Node Shipping
- Thumbnails, waveforms, and catalogue indexing through Media Nexus Shipping
- AI-generated descriptive metadata and tagging Early access
- Visual understanding and semantic search across the library Planned
Move your media
- Watch-folder ingest with stable-file detection Shipping
- Storage role enforcement — sources stay read-only Shipping
- Backup Guard checks against declared archive locations Shipping
- Approval-gated moves and copies between locations Shipping
- Connections to Frame.io, Drive, Dropbox, S3, Slack, and email Early access
- Automated delivery and upload workflows Early access
- Checksum verification across full ingest and archive cycles Planned
- Dispatching work to edit bays and other endpoints Planned
Automate your operation
- Durable workflows: trigger → steps → approval → result Shipping
- Manual, new-media, scheduled, and conditional triggers Shipping
- A job queue that survives crashes and reboots Shipping
- Persistent background workers on a schedule Shipping
- Alerts for lost storage, failed jobs, and missing credentials Shipping
- An approval inbox naming the exact capability and scope Shipping
- A full activity history of every job, run, and approval Shipping
Stay in control of it
- A local control center — health, jobs, workflows, logs Shipping
- Pause or resume any worker without touching a terminal Shipping
- Per-Node and per-worker spend ceilings on AI usage Shipping
- Explicit grants for every AI provider and class of data Shipping
- Vaultline support actions that are typed, audited, and refusable Shipping
- Self-managed mode with no control plane at all Shipping
- Fleet view across multiple Nodes and sites Planned
How Media Node works
Media Nexus knows what exists. Media Node lets it act.
Knowing your media and acting on it are different problems. A Node is three things stacked on top of each other to close that gap — and only the third one is the reason it's worth what it costs.
01
Dedicated infrastructure
An always-on machine, local to your storage, so processing isn't waiting on a laptop that's asleep. Necessary — but on its own, this is not why a Node costs thousands of dollars.
02
The Vaultline AI Router
One place that decides where a task runs — local or cloud, by an explicit grant and a spend ceiling — instead of a customer wiring together five separate AI products themselves.
03
Media agents & tools
The model doesn't get the keys to the machine. It invokes tools Vaultline has built and tested, in a sequence it decides, inside limits it cannot exceed. This is the part we're actively building toward.
A normal Media Nexus installation knows what exists — the files, the metadata, the projects, the relationships between them. Media Nexus with a Node can reason, decide, execute, verify, and continue against that environment. That's the leap a Node is actually for.
What's running on the Node todayGoverned autonomy
ShippingEvery capability sits in a tier: observe runs on its own, report writes only to Node-owned state, modify stops for a named approval, and destructive operations have no runtime capability at all — not a permission that's off, a code path that doesn't exist.
Persistent AI agents
ShippingResearcher, Librarian, and Operator run continuously against your library. They're standing workers with a schedule and an allowance, not a chat window you have to prompt.
Connector Vault
Early accessDirect, narrowly-scoped connections to Frame.io, Drive, Dropbox, S3, Slack, and email — each with its own credential and a fixed set of allowed actions, commissioned from the Control Center without opening a terminal.
Spend governance
ShippingNode-wide and per-agent ceilings on AI usage, checked before any request leaves for a cloud provider, with a running ledger of what was spent and on what.
A queue built for what's next
PlannedEndpoint policy and a durable, typed task queue for dispatching work to edit bays and other distributed endpoints. The queue ships; dispatch adapters are next.
This is the substrate pillar three runs on. What we're building on top of it is a growing library of tested, deterministic tools an agent can invoke — search, transcode, transcribe, file operations, delivery — each one software we've written and verified, not a model with a shell.
Pillar 02 · The AI Router
One place that decides where a task runs.
A customer shouldn't have to know whether a job needs a local model or a frontier one, or wire together five separate AI products to find out. The Node makes that decision once, in one place.
- Routing is local-first. Media-heavy work — transcode, large-batch transcription — stays on the Node. It doesn't need to leave, and it's cheaper not to move it.
- Text, reasoning, and agent workloads can burst to Vaultline Compute — Vaultline's managed cloud capacity, metered against your monthly pool — when a task calls for more than local models can do.
- Anything bound for the cloud needs an explicit grant naming the class of data it may see. Nothing unnamed gets through.
- Every request is metered against a spend ceiling you set, per Node and per worker, checked before it leaves the building.
Transcode & transcriptionlocal only
Descriptive metadatalocal runtime
Reasoning & agent tasksVaultline Compute
Unnamed providerrefused
Monthly allowance
Vaultline Compute, enforced before the request leaves
Within ceiling
local Paths, filenames, and transcript text stay on the Node.
compute Only the granted data class, metered against your pool.
How the pieces fit
Media Nexus gets them in. Media Node deepens the relationship.
Vaultline Managed is what makes the whole stack — infrastructure, Nexus, and the AI layer — actively managed and getting more capable over time, instead of a box you bought once and left alone.
The intelligence layer
The catalogue, the search, the workflows, the policy, and the interface your team actually opens. It holds the record of what your media is and what should happen to it.
- Installable and ownable on your own hardware, on its own
- Ships running on the Node from day one, if that's the path in
The permanent presence
The machine inside your building that hosts Nexus, reaches your storage, and executes the work — plus its own workflow engine, agents, and governance layer that keep running whether or not Nexus is open. See what that adds.
- Gives Nexus persistent local compute
- Also runs its own automation independent of Nexus
Media Nexus stands on its own. Install it on your own hardware and it's a complete system — catalogue, search, workflows, nothing missing. Media Node is the upgrade you reach for when you want dedicated infrastructure, a managed AI layer, agentic workflows, and a fully managed Nexus environment, all under one membership. Not a bundle you're pushed into on day one — a step you take when reliability and capability are worth more than running it yourself.
If you already run Media Nexus, the Node becomes the machine it lives on and nothing about your catalogue changes. If you're starting fresh, day one of the Node is day one of Nexus.
Deployment
Configured by us. Activated in your environment.
We preconfigure and test the Node before it ships. Your team connects power, Ethernet, and storage. We remotely complete activation, integration, and the first workflows.
01
We build and test your Node
Hardware provisioned, software installed and signed, and an acceptance run passed before it ships — including reboot recovery and a real media job end to end.
Vaultline
02
Plug it into your environment
Power and Ethernet, next to your storage. That is the whole on-site task — no monitor, no keyboard, no installation session.
Your team
03
We activate it
We connect your storage volumes, stand up Media Nexus, wire your integrations and credentials, and build the first workflows around how your team already works.
Vaultline, with you
04
It runs
From then on it is infrastructure. It watches, processes, indexes, and reports — and when something needs a human decision, it stops and asks rather than guessing.
OngoingSupport is optional and explicit. Remote support is off by default, every supported action is a named operation rather than a shell, and you can disable it from the control center at any time.
The appliance
An appliance, not a computer you have to think about.
Delivered on Apple silicon in a chassis small enough to sit behind the NAS. What matters isn't the SKU — it's that it is quiet, efficient, and built to run untouched.
- Compute
- Apple silicon, sized to your media volume and AI workload
- Memory
- High unified memory for concurrent media and model work
- Network
- 10 Gigabit Ethernet where your storage supports it
- Footprint
- Desktop-sized, quiet, low draw, designed for 24/7 operation
- Storage
- Internal working storage; your media stays on your volumes
- Management
- Signed releases, health-gated updates, and rollback
Delivered as one unit, not a parts list.
You do not spec it, order it, image it, or patch it. The Node arrives built and tested, it updates itself from signed releases with a health check and a rollback path, and if it ever has to be replaced the configuration comes back with it.
- Sized by us against your storage and workload
- Signed, notarized releases with health-gated updates
- Rollback if an update misbehaves
- Runs headless — no monitor, no keyboard, no login
Configurations are set per deployment, after we look at your storage, your volume of media, and the workflows you want running. We'd rather size it correctly than publish a spec sheet you have to work around.
In practice
What it looks like on a normal Tuesday.
The same appliance, pointed at three different operations.
Cards offloaded at 11pm
A DIT dumps the day's cards to the NAS and goes home. By the morning standup the footage is registered, proxied, transcribed, and searchable — nobody stayed late to kick it off.
Footage lands → Node detects → metadata → proxies → transcript → indexed
Nine years of campaigns
Shared storage is full of work nobody can find. The Node grinds through the backlog in the background and keeps up with new arrivals, so "do we have a shot of that?" has an answer.
Shared storage → continuous indexing → enrichment → search
Client media, every week
Deliveries arrive from a dozen sources in a dozen shapes. The Node normalizes them into the same structure, pushes what needs reviewing, and runs archive on a schedule.
Client delivery → organized → processed → review → archived
Built to be trusted
Boring on purpose, underneath.
The Node is assembled from tools post-production engineers already trust — wrapped in the operational discipline an appliance requires.
In the stack
FFmpeg
ffprobe
whisper.cpp
ExifTool
rsync
SQLite
launchd
SMB / NFS
Apple silicon
Local model runtime
No inbound ports
The Node opens nothing to the internet. It reaches out on one authenticated connection and accepts only the typed actions your policy allows.
Sources stay read-only
Original media is mounted so automation cannot write to it. Anything that would modify or move media stops for a named human approval.
Signed, gated releases
Updates are signed and notarized, applied behind a health check, and rolled back automatically if the Node comes up unhealthy.
Everything on the record
Every job, worker run, approval, and support action lands in a durable activity history you can read from the control center.
Pricing
Buy the appliance once. Vaultline Managed keeps it current.
The Node is a one-time purchase — you own the hardware outright. Vaultline Managed is what keeps the infrastructure, Media Nexus, and the AI layer actively maintained and improving, with the first three months included.
Media Node
For a working media library.
$6,000
One-time · hardware included · yours to keep
- The appliance itself, built and tested before it ships
- Storage connection, integrations, and credentials wired up
- Your first workflows configured around how your team works
- Remote activation and handoff
- Three months of Vaultline Managed — Media Nexus included
Comfortable with
Nightly card dumpsProxies + transcripts
Continuous indexingScheduled workflows
Media Node Pro
For queues that never empty.
$8,500
One-time · hardware included · yours to keep
- Everything in Media Node
- Multiple proxy and transcription lanes running at once
- Larger local AI models without starving media jobs
- Built for multi-terabyte ingest days and deep back-catalogue runs
- Three months of Vaultline Managed — Media Nexus included
Reaches for
Daily multi-TB ingestConcurrent lanes
Years of backlog
Choose by workload, not by spec sheet: if one day's footage regularly outruns overnight processing, or you want the backlog and the AI enrichment moving at the same time, you want the Pro. We'll tell you which one fits when we look at your volume.
Vaultline Managed · first three months included
$475 / month
Keep your entire Vaultline system current, managed, and getting more capable over time — not a fee for the box existing.
Managed infrastructure
- Health monitoring and Node status
- Service recovery and updates
- Remote diagnostics
Managed Media Nexus
- Module updates
- New codec and file-format support
- New integrations, including Relink expansion
Managed AI layer
- Model and routing updates
- New agent tools as they ship
- Local-model updates and frontier-model routing
Included usage
- A monthly pool of AI usage
- A monthly pool of Vaultline Compute
- Transparent, metered overage if you go past it
Support covers bug fixes, configuration changes, and minor adjustments to existing workflows. New custom workflows or integrations are scoped separately.
Already a Partner member? Vaultline Managed is included — no separate charge.
The whole picture
Two numbers, no assembly required.
Once
Media Node or Node Pro — the appliance, configuration, activation, and your first workflows.
$6,000 or $8,500
Monthly
Vaultline Managed — infrastructure, Media Nexus, and the AI layer kept current, plus a monthly pool of AI usage and Vaultline Compute. First three months included.
$475 / mo after month threeThat is the entire cost of the system. If a Partner membership already covers you, the monthly line disappears.
Questions
The things teams ask first.
Is this just a Mac mini you configure for us?
The hardware is how the infrastructure gets delivered, not what you are buying. What you are buying is an always-on automation layer for your media operation: Media Nexus, the job queue, the workers, the workflow runtime, the safety boundary, and the deployment and support around it.
Does it replace our NAS?
No. Your media stays exactly where it is. The Node mounts your existing storage — sources read-only — and writes derivatives only to locations you designate for that purpose.
What can Vaultline see or do on our Node?
Health and status, not your media. Paths, filenames, transcript text, and raw diagnostics stay local. Remote support is a separate grant that is off by default, every supported action is a specific named operation rather than a shell, and you can disable it from the control center.
Can it run without a connection to Vaultline?
Yes. Self-managed mode has no control plane, forbids Vaultline remote support, and rejects Vaultline-managed AI routing. Same binary, same local safety boundary.
Do we need Media Nexus too?
Not necessarily — Media Nexus can be installed and owned on your own hardware as a complete system on its own. If a Node is your path in, it ships with Media Nexus already running on it, and Vaultline Managed keeps both current together. If you already run Media Nexus, the Node simply becomes the machine it lives on.
What does it cost?
$6,000 for a Media Node, $8,500 for a Media Node Pro — one-time, hardware included, with the configuration work to connect your storage and stand up your first workflows. Vaultline Managed is $475 a month after the first three included months: managed infrastructure, a managed Media Nexus, a managed AI layer, and a monthly pool of AI usage and Vaultline Compute. Support covers bug fixes, configuration changes, and minor adjustments to existing workflows — new custom workflows or integrations are scoped separately. Partner members have it bundled already.
What happens if we stop Vaultline Managed?
You keep the Node — you bought it — and the workflows configured on it keep running. What rides on Vaultline Managed is everything that keeps improving: Media Nexus and AI-layer updates, monitoring and signed updates, included usage, and our help changing things. We'll walk through what that means for your setup before you decide, not after.
Next step
Build a media operation that keeps working when your team goes home.
Tell us about your storage, your volume, and the workflows you are doing by hand. We will tell you honestly whether a Node is the right answer yet.
$6,000, hardware included
Vaultline Managed included
Commissioned individually