Tubeup 2.0: YouTube Automation That Works Like an Assistant

Ask a YouTube creator where their week goes, and the honest answer is rarely "making videos." It goes to the part after the video: typing titles, pasting the same link block into forty descriptions, choosing playlists, setting visibility, picking a publish time, switching to a client's channel, and doing it all again. A lot of that happens late at night, because that's when the edits are finally done.

Tubeup has always been about getting that time back. Version 1.0 made uploading fast. Tubeup 2.0 takes you out of the upload session altogether. The shortest way to describe it: point Tubeup at a folder, and it writes the metadata, uploads on your schedule, drafts the social posts, and tells you what it did.

This post isn't a changelog. It walks through the problems creators and agencies actually have and how each part of Tubeup 2.0 solves them, so you can tell whether it fits the way you publish.

The problem: publishing is a second job

If you publish more than a couple of videos a week, the work around each upload adds up to something bigger than any single task:

• Repetition. The same description skeleton, the same sponsor links, the same hashtags, retyped or pasted per video.
• Presence. Uploads that only happen while you're sitting at the computer, so publishing is tied to your working hours.
• Channel juggling. Agencies and multi-channel owners switching accounts and remembering which video goes where.
• AI bills. Metadata generation that gets more expensive the more you publish.
• Fragility. Everything lives in one place, and one bad reinstall or dead laptop takes your whole setup with it.

Tubeup 2.0 has an answer for each of these. They add up to more than the sum of the features.

Automation: set it once, and it publishes without you

The biggest change in 2.0 is automation flows. A flow runs Tubeup's own pipeline on a loop — watch a folder, write the metadata, upload, promote — and each step does exactly what its button on the Dashboard does. Nothing is hidden inside a black box, so a flow behaves the same way you would.

Every flow gets:

• Its own schedule, so a channel publishes at 6pm daily or three times a week without you there.
• A daily cap, so automation can't run through your upload quota.
• An activity log and a backlog view, so nothing happens off the record and you can see what's queued next.
• A Simulate button that shows the whole run, with the exact videos, metadata and destinations, before a single video moves.

The detail that matters most for trust: when a flow pauses, it tells you why. It won't just say "flow stopped". It says quota exceeded, this channel is no longer connected, or three failures in a row. You never come back to a dead automation with no idea what happened.

In practice, your upload session becomes a folder. Drop fifty edited videos into it tonight, and the flow publishes them on the cadence you set, overnight, over a weekend, or while you're shooting the next batch, with a log of every one.

Chat with Tubeup: stop clicking, start typing

The second big change is a chat dock that opens over any page: Dashboard, Uploaded, Analytics or Settings. You can ask it about your library and it answers. More usefully, you can tell it what to do.

"Set every video in this batch to Unlisted and schedule them for 6pm daily starting Monday" is one sentence instead of forty clicks. Bulk edits stop being a chore with a scrollbar.

A personal publishing assistant, with you in charge

The chat isn't a help bot that points you to menus. It can take 14 real actions across the whole publishing workflow:

• set a schedule, set visibility, set a playlist, or edit fields
• generate metadata or generate social copy
• upload, cancel an upload, or delete videos
• apply a content profile or edit one
• check automation status, pause automation, or enlist new videos

That's what makes it feel like handing instructions to an assistant. You can say "generate metadata for everything I added today and put it in the Tutorials playlist", or "pause the automation", or "add my new Discord link to the Sponsor profile".

What keeps it safe is the part we'd never cut: every action is shown as a diff and waits for your confirmation. The model picks an action from a fixed list and never writes to your library directly. Nothing runs until you say so. If you've been wary of letting AI anywhere near your channel, this is the design that should settle it: the assistant proposes, and you decide.

Better channel integration: every channel you own, one Google project

For anyone running more than one channel, this was the feature people were waiting for.

In Tubeup 1.0, the API engine could hold one channel; connecting a second replaced the first. In 2.0, the API engine connects as many channels as you own from a single Google Cloud project, with one client ID and secret. Press Authorize once per channel and pick that channel on Google's consent screen. If one channel needs its own quota, give it its own per-channel credentials on a separate Cloud project.

Around that, the channel experience got the polish it needed:

• One channel list across both upload engines, showing whether each channel is ready for Browser uploads, API uploads, or both.
• A channel filter in Analytics, to switch between every connected channel without leaving the page.
• A confirmation before removing a channel, where 1.0 took a single click.
• Removing the default channel picks a new one, so uploads are never left pointing at a channel that's gone, and renaming a channel keeps its flows working.

For an agency, the pitch fits in one line: ten client channels, one Google project, one app.

Per-video, per-channel uploads

Connecting every channel only helps if each video goes to the right one. In 2.0 you pick the destination channel on the video row itself. A mixed batch — three videos for one client, five for another, two for your own channel — goes to the right places in a single pass.

The same thinking applies to timing. The schedule picker opens right where the video is and refuses a time in the past, so an impossible date is caught when you type it, not when the upload runs.

Content profiles: your brand's description, written once

This is the single biggest cut to repetitive typing in the app.

A content profile is a named, reusable set of the blocks that make up a description: brand links, sponsor links, suggested links, timestamps, resource links, social links, hashtags and SEO tags, plus its own label visibility. Think of it as a brand's publishing identity, saved.

You can use profiles in several ways:

• Bound to a channel, so every video on that channel carries the right links and tags automatically.
• Applied to a batch in one action.
• Applied by a flow, so automated uploads get the right identity without you.
• Edited from chat. "Add my new Discord link to the Sponsor profile" updates it everywhere it's used.

For a creator with several channels, or an agency with several clients, this is where the time goes back. A sponsor set for one channel, an affiliate set for another, a clean set for a third, and you switch the whole description body by picking a name. Before, you pasted your link block fifty times. Now you pick a profile.

If you're upgrading from 1.0, your existing defaults become a permanent Global Default profile on first launch, so your descriptions publish byte-identical on day one.

A faster publishing workflow, end to end

Put those together and the day-to-day workflow changes shape:

1. Drop videos in a watch folder. Enlisting thousands of files writes in batches, so large drops don't stall the app.
2. Let a flow write the metadata, apply the right content profile, and pick up the channel you set.
3. Simulate once to check the run, then let it publish on schedule.
4. Ask the chat for any exception, like "move Tuesday's video to Friday", and confirm the diff.
5. Check the Uploaded page, a read-only record of exactly what each video was uploaded with. You no longer need to open YouTube Studio to see what went out.

The app itself also got faster, which matters when you leave it running. The backend now starts in 0.56 seconds instead of 1.44, uses 53 MB of memory instead of 154, and with 5,000 videos in the library the list answers in about 100 ms, with typing costing about 3 ms per keystroke. The interface got denser too, with 48px rows that fit more videos on screen, a searchable Settings page, and clearer tables across every view.

More powerful AI, with no lock-in

Tubeup 1.0 used Google Gemini. Tubeup 2.0 supports nine AI providers: Revoye, Google Gemini, Anthropic, OpenAI, DeepSeek, Qwen (DashScope), Perplexity, OpenRouter, and any OpenAI-compatible server. It also runs local models through Ollama or LM Studio, for creators who want nothing leaving the machine.

A few details make this more than a longer dropdown:

• Every provider's model is editable, because vendors retire models on their own schedule and you shouldn't be stranded when they do.
• Switching providers keeps every other setting.
• You decide how many videos each AI request carries, so metadata generation fits your provider's limits.
• AI is woven through the workflow: metadata, social copy, flows, and the chat assistant itself.
• Existing installs keep the provider they already use. Nothing is switched under you.

Revoye: the AI provider we recommend for Tubeup

Revoye is Tubeup 2.0's new default AI provider, first in every picker, and it's the one we recommend.

Most AI providers bill by the token, so the more videos you publish, the bigger the AI bill. Metadata for a 200-video batch used to mean an API cost that grew with your output. Revoye works differently. It runs prompts through the AI accounts you're already signed into, the subscriptions you already pay for, so there is no per-token or per-word charge.

For Tubeup users, that means:

• Zero additional AI token cost on the default path. You pay nothing extra per word, however many titles and descriptions you generate.
• No token metering to watch. Generating metadata for 20 videos or 200 costs the same on the AI side, so you don't have to ration batch generations. The only limits are the ones your own AI subscriptions already have.
• A provider built for this kind of work. Revoye is made by the same team as Tubeup, and it's designed to turn the AI subscriptions you already own into something your apps can call.

To be precise about what Revoye is: it doesn't resell anyone's model, and it isn't an OpenAI or Anthropic product. It routes your prompts through the AI accounts you've signed into yourself. You can read how it works in Your AI Subscriptions, as an API, or visit the Revoye product page. Its dashboard lives at revoye.com.

And if you'd rather not use it, you don't have to. Any of the other eight providers, or a local model, works the same way.

Full data backup and restore

Here's a question worth asking about any tool you've invested hours of setup into: what happens when your laptop dies?

Tubeup 1.0 backed up your settings. Tubeup 2.0 backs up your work. Every video row, setting, content profile, flow and chat thread goes into one file. You can restore it two ways:

• Merge: restore over what's there, combining the backup with your current library.
• Replace: restore in place of it, for a clean start from a known state.

That covers the situations that actually happen: moving to a new machine, reinstalling, or undoing an experiment that went wrong. It's also built to be safe. A restored backup can't switch automation on, can't enable a flow that fails validation, and can't point your library at a folder that doesn't exist. A backup file is data, not a command.

Your data stays yours throughout. Tubeup is a desktop app with a local database; there is no Tubeup cloud holding your content. Tubeup 2.0 also locked down its local backend, so web pages open in your browser can't reach it.

How it all works together

Individually these are good features. The point of 2.0 is how they connect:

• A flow picks up new videos from a folder.
• Revoye writes the metadata at no per-word cost.
• The right content profile stamps each video with the right brand's links and tags.
• Each video goes to the channel chosen on its row, across as many channels as you own.
• It publishes on schedule, within a daily cap, with a log of every run.
• When something needs changing, you ask the chat and confirm the diff.
• And a backup holds the whole setup in one file.

That's the difference between a faster uploader and a publishing workflow that keeps going without you.

Who gets the most out of it

• Solo creators who spend more time uploading than creating. Your upload session becomes a folder you drop files into.
• Content agencies running client channels, with one Google project, per-video channel selection, and a content profile for each client.
• Faceless and high-volume channels publishing hundreds of videos, using flows with caps, AI with no per-word bill, and an app that stays fast at 5,000 videos.
• Beginners who aren't sure what automation will do: press Simulate and watch the whole run before anything moves, or just ask the chat.

If you're coming from manual uploads, our earlier guide to bulk uploading and automating YouTube explains the basics. If you've hit YouTube's API limits, why "quota exceeded" happens explains the limit, and why Tubeup has two upload engines.

Try Tubeup 2.0

Tubeup runs on Windows and macOS. Every feature above is available in a 3-day free trial with no card, and every plan is a one-time lifetime licence with all future updates included. Plans differ only by how many computers you install on. See the plans and start the trial on the Tubeup product page.