Growhub Media Request early access
How it works One filesystem AI agents Workflows FAQ Request early access

AI-native cloud filesystem Early access

Your files live in the cloud. Your workflow stays local.

Growhub Media mounts cloud storage as a drive on your computer. Browse terabytes of footage, images, and project files in ordinary folders, open them in the applications you already use, and retrieve only the data those applications actually read.

Request early access See how it works
  • Mounted like a local drive
  • Streamed on demand
  • Cached where it helps
Interface preview — illustrative, not a final product screenshot.
01 — The problem

Libraries grew. Workflows didn't.

A single documentary can produce hundreds of terabytes of camera originals. A studio's archive, a game team's assets, a training dataset — they outgrew the drive on the desk long ago. Yet most cloud storage still asks you to download before you can do anything.

Download-first

  1. Find the file in a web app
  2. Download 184 GB
  3. Wait
  4. Open it
  5. Upload your changes

With Growhub Media

  1. Open northline-doc/footage/A001_C004.mov
  2. Your app reads the frames it needs
  3. Growhub retrieves those ranges
  4. Keep working
02 — How it works

The cloud, mounted.

Your entire library, without your entire library on disk.

  1. 01

    Mount

    Growhub Media appears on your computer as a drive with ordinary folders and paths. Your file browser and applications see files — not a sync queue.

  2. 02

    Open

    When an application reads a file, Growhub Media retrieves the parts it asks for from cloud storage. An editor jumping to minute forty doesn't wait for the first thirty-nine.

  3. 03

    Cache

    Data you've recently used can be kept in a local cache, so repeat reads don't travel back to the cloud — without turning your disk into a second copy of the library.

A001_C004_0412.mov Retrieved 0.0 of 184.2 GB
Illustrative: an editor plays the opening of a 184.2 GB camera file, then jumps ahead. Only the ranges the application reads are retrieved — the rest stays in the cloud.

What happens when you open a file

  1. Your application Opens a file at a path, exactly as it would on a local disk.
  2. Growhub drive The mounted filesystem receives the read request.
  3. Local cache Serves data already on this device.
  4. Cloud storage Holds the full library and returns only the requested ranges.
Different by design

Not a place to upload files. A place to work with them.

Conventional cloud storage Growhub Media
How you reach files A web app, a sync client, or a bucket API A mounted drive with ordinary paths
Before you can start Download or sync what you need Open the file; data is retrieved as it's read
Local disk needed Roughly the size of what you sync Only what's cached
Who works from it Mostly people, each with their own copy People, applications, and supported AI agents — one namespace
Copies created One per person, machine, or pipeline One library, accessed in place
03 — The bigger idea

One filesystem for people, applications, and AI agents.

Storage usually ends with the person who uploaded the file. Growhub Media is designed as a persistent namespace — the same paths, the same files — that authorized people, devices, applications, and agents can all work from.

  • Editor, LisbonMounted drive
  • Producer, on the roadMounted drive · second device

Workspace · northline-doc

  • footage/412 files
  • audio/
  • transcripts/
  • NL_edit_v12.prproj
  • selects/
  • deliverables/
  • Transcode scriptFilesystem paths
  • AI agentAgent interface Planned
04 — AI-native Planned capability

Agents that work where the files are.

Handing media to an AI tool today means exporting, uploading, and re-importing whatever it produces. Growhub Media is being designed so supported agents can discover, read, process, and write files through filesystem interfaces, APIs, or SDKs — inside the permissions you grant, in the workspace your team already uses.

  • DiscoverList and search paths instead of being handed a zip file.
  • ReadRetrieve only the files and ranges a task needs.
  • Write backResults land as ordinary files your team can open and review.
  • Stay in scopeAccess limited to what you grant, with changes attributable to the agent that made them.
agent session · selects-assistant
scope /northline-docread /northline-doc/selectswrite
  1. list /northline-doc/footage --type video 412 files
  2. read /northline-doc/transcripts/interview_03.srt 38 KB
  3. read /northline-doc/footage/B014_C002.mov 00:12:40 → 00:14:05
  4. write /northline-doc/selects/interview_03_selects.edl created
  5. Selects ready for review in /northline-doc/selects
Illustrative — the agent interface is planned and not yet available.
05 — Workflows

Stop moving files. Start working with them.

Edit from the full archive, not the slice you synced.

Camera originals, proxies, audio, and project files live in one workspace. Editors open what the timeline needs; assistants and colorists work from the same paths instead of couriered drives.

  • Scrub large media without syncing the whole card dump first
  • Keep media links pointing at stable paths across machines
  • Bring review and finishing onto the same library
  • northline-doc/footage/61.8 TB
  • A001_C004_0412.mov184.2 GB
  • northline-doc/proxies/1.2 TB
  • northline-doc/audio/212 GB
  • northline-doc/edit/NL_edit_v12.prproj
For developers

It's a filesystem. Your code already speaks it.

Anything that reads files through the operating system can read from the mounted drive. A dedicated API and SDK for workspaces and agents is planned.

Standard file APIsMount path
from pathlib import Path

footage = Path("/Volumes/Growhub/northline-doc/footage")

for clip in sorted(footage.glob("*.mov")):
    with clip.open("rb") as f:
        f.seek(4 * 1024**3)        # jump 4 GB in
        chunk = f.read(1 << 20)   # read 1 MB from there
Growhub SDKPlanned · illustrative
// API design in progress — package not yet published
import { Growhub } from "@growhub/sdk";

const ws = await Growhub.workspace("northline-doc");

for await (const file of ws.list("footage", { type: "video" })) {
  console.log(file.path, file.size);
}

await ws.write("selects/notes.md", summary);
Where we are

What early access covers, and what comes next.

We'd rather be precise than impressive. Growhub Media is in early development, and early access focuses on the core filesystem experience. Everything marked planned is on our roadmap — plans can change, and we'll say so when they do.

Early access focus

  • Mounted driveFamiliar folders, paths, and file operations on your computer.
  • On-demand streamingFile data retrieved when applications read it.
  • Local cachingRecently used data kept close for repeat access.
  • Shared workspacesOne workspace mounted across people and devices.

Planned

  • AI agent interfacesScoped discovery, reading, and writing for supported agents.
  • Developer API & SDKProgrammatic workspace access for apps and automation.
  • Version history & recoveryRestore earlier versions of files and folders.
  • Drive forksIsolated project versions you can branch and compare.
  • Admin & deployment optionsOrganization controls and choices for where data is stored.
Questions

Straight answers.

Including the limitations.

What exactly is Growhub Media?

A cloud filesystem. It mounts on your computer like a drive, but the files live in cloud storage and their data is retrieved when applications read it. The longer-term goal is one persistent filesystem that people, applications, teams, developers, and supported AI agents can all work from.

How is it different from a sync client or a storage bucket?

Sync clients copy files to each device before you work with them, and buckets are reached through APIs or web consoles. Growhub Media presents your library as a drive with normal paths and retrieves file data as it's read — so you work with files in place rather than moving them around first.

How can I work with terabytes without downloading everything?

Applications rarely read a whole file at once. A video editor reads the frames on screen; a script reads the records it's processing. Growhub Media retrieves those specific ranges from the cloud and can keep recently used data in a local cache, so local disk use tracks what you're actually working on.

Will it work with all of my applications?

The goal is to work with applications that read and write files through the operating system. Some apps make assumptions about local disk speed or file locking, so we're testing media and creative tools during early access and will publish what we've verified rather than promise universal compatibility.

What happens when I'm offline?

Files, or parts of files, that aren't cached locally need a network connection to open. We don't claim offline support beyond whatever is already in your cache.

How fast is it?

It depends on your connection, your distance from storage, and how your application reads files — playing through a clip behaves differently from jumping around uncached data. We'll share real measurements from early access instead of headline numbers.

What makes it AI-native?

AI agents need to find, read, and produce files just like people do. Instead of exporting data to a separate AI tool, Growhub Media is being designed so supported agents can work in the same workspace through filesystem interfaces, APIs, or SDKs, within the access you grant. Agent interfaces are planned, not yet available.

Where are my files stored, and who can access them?

Files are kept in cloud object storage and reached only through the access granted to each person, device, or application. Admin controls and storage deployment choices are planned. We'll publish our security practices as they're finalized, and we don't currently claim any security certifications.

How much does it cost?

Pricing hasn't been set. Early-access participants will hear about it first, well before anything is billed.

06 — Early access

Bring your biggest library. We'll bring the drive.

Early access is opening in small groups, starting with media-heavy teams. Tell us what you work with and we'll be in touch as places open.

Building pipelines or agents? Ask about developer access

Submitting opens your email app with these details filled in. Nothing is stored by this website.