Set Up a Pipeline Once and Stop Opening New Campaign

Everything you publish still starts the same way: you open the app, click New Campaign, paste something in, pick the channels, and wait. It works. It also means nothing happens on the days you don't sit down and do it.

So the good stuff piles up somewhere else. A post you wrote on Facebook that never made it to LinkedIn. A folder of screenshots from last week's release that everyone agreed should go out. A changelog entry sitting in a repo. None of it is blocked on writing — it is blocked on you being in front of the app at the right moment, with enough attention left to start the flow.

What changed

There is now a Pipelines view that watches a source for you and turns whatever shows up into channel-ready content on a schedule — either a channel you already post to, or folders on your own machine. You configure it once and it keeps running; new posts get adapted, new files become campaigns, and the results land as drafts or scheduled posts without you opening New Campaign at all.

How it works in practice

Point a pipeline at a channel and let new posts fan out

A Social pipeline watches one source account and repurposes each new post for the destinations you pick. Configuration is four tabs — Source, Transform, Destinations, Delivery — and they reuse the same options you already set in New Campaign: language, how much detail to keep, output format, and whether pieces land as drafts or scheduled posts.

Connected-account sources cover Facebook Pages, X, Instagram, Threads, YouTube, TikTok, Bluesky and Mastodon. You choose how often it looks — Every hour through Every week — and the pipeline only acts on posts newer than the milestone you set when you saved it, so turning it on doesn't dump your entire back catalogue into the queue.

You can also hit Run now whenever you want it to go immediately rather than waiting for the next check.

Watch a public URL when the channel won't hand over a list

Some channels simply won't list your posts through an API — a personal Facebook profile, LinkedIn without read approval. Instead of leaving those out, a pipeline can poll a public profile or post URL and trigger when the latest readable post changes.

It is a best-effort HTML fetch, and it says so: private or login-walled pages fail with a clear message instead of quietly returning nothing forever. The first scheduled check records a baseline, so only genuinely new posts fire a run.

Drop a file in a folder and get a post out

The Directories tab is a folder pipeline. Point it at folders on your machine and every new file that lands there becomes a campaign on the next check — no paste step, no upload step.

You pick which file families count — Markdown (.md, .mdx, .markdown), text and data (.txt, .csv, .json), images, video — plus any extra extensions you name. A shot.md sitting next to shot.png isn't treated as its own post; it's read as that image's caption brief. And you can add notes describing what lives in those folders and how it should be framed, which apply to every file the pipeline picks up.

Two guards matter here. Ignore files that are already there is on by default, so pointing a pipeline at a folder holding 400 files does not generate 400 campaigns — turn it off deliberately when you want a first run to sweep what's already on disk. And media is copied into app storage as it's picked up, so a scheduled post survives you moving or deleting the original file.

Decide whether a folder is one post or many

A folder of material is sometimes one story and sometimes ten. Both are supported explicitly.

Every file is a post turns each matching file into its own campaign. Every subfolder is a post bundles everything inside a subfolder into one campaign — text becomes the brief, images and video get attached — while loose files in the watched folder stay one post each. A bundle waits until every file in it has finished copying, so a folder still being written never publishes half of itself.

A live preview in the config shows exactly which files and bundles would be picked up before you save.

Override one channel without forking the pipeline

Inside a folder pipeline, each selected destination shows a summary strip — EN · Plain text · Balanced — that opens the same inline settings panel used in New Campaign. One channel can run in two languages, in Markdown, at a more detailed length, while the rest keep the pipeline defaults.

Let the writer look things up

Separately from pipelines, there is now a Web switch in the campaign composer. Turn it on and the content AI can call web search while writing, instead of working only from your brief and whatever the model already knows.

It is off by default, which is the right default — ordinary runs stay cheaper and work offline. Turn it on when you need current facts, competitor context, or citations. Hosted search works with your own OpenAI, Anthropic and Gemini keys, and local CLI agents that expose web tools — Grok among them — can use theirs.

You also get to see what it looked at. When search runs, Generation Logs open and list each query and then the results as concrete sites — hostname, title where available, and a link you can open. Header badges count queries and sites as the run goes. The pre-flight token estimate flags it too, noting that hosted search adds latency and provider fees on top of the token figure.

Before vs after

Getting a post onto other channels	Before	Now
Noticing there's something new to post	You remember, eventually	The pipeline checks on your interval
Starting the work	Open New Campaign, paste the source	Already running
Turning a folder of assets into content	Upload each file by hand	Drop it in a watched folder
A first run on an existing folder	Not applicable	Baselined by default, opt in to sweep
Knowing which pieces were automated	No way to tell	Pipeline badge and filter in Distribution History
Writing with current facts	Paste them in yourself	Web switch, with the sources logged

Who benefits most

If you post in one place and mean to post in five. A social pipeline pointed at the channel you actually use turns that channel into the source for the rest, on a schedule, without a second writing session.

If your content starts as files. Screenshots, release notes, exports, recordings — anything that already lands in a folder as part of your work becomes a campaign without a separate "now market it" step.

If you're the only one doing this. The run log records what was fetched, generated and scheduled, with the source title and any error, so a quiet failure is readable after the fact instead of being a week of silence you notice too late.

Try it

Open Pipelines in a project, pick Social or Directories, and configure one — a channel you already post to, or a single folder you drop work into. Set the interval, choose whether output lands as drafts or scheduled posts, and save. Then jump from a run row in the Logs tab straight to the campaign it produced, or to Distribution History filtered to that pipeline.

Set it up once and the queue keeps filling on the days you don't open the app.