Source automation
Register a news source and the system scans for new articles, renders video and publishes on schedule. This is where the pipeline runs without you.
Registering a source
- Paste the publication URL or RSS feed you want to follow.
- Give it a memorable display name for when you have several.
- Choose a branding template, a voice, default hashtags and background music.
- Choose which channels receive the automated videos.
The first scan runs immediately on save
The source checks for new articles the moment you save, rather than waiting for the next scheduled slot. That way you find out straight away whether it works.
To upload your own music you must save the source first, then reopen it (mp3/m4a ≤ 5 MB, up to 5 tracks).
The scan schedule
Four presets fill everything in quickly. Editing any field below switches to custom.
| Preset | Configuration |
|---|---|
| Breaking news | 06:00–22:00, every 30 minutes, 2 newest articles |
| Three times a day | 07:00 · 12:00 · 19:00, 3 articles each |
| Drip | 06:00–22:00, every 3 hours, 1 random article each |
| Custom | Everything set by hand |
Two scan modes:
- Even interval — repeats within the window you set, e.g. every 2 hours between 07:00 and 22:00.
- Fixed times — scans only at the hours you pick, identically every day.
With no window set, the source runs around the clock
Leaving the start and stop hours blank means it scans at 2 and 3 in the morning too. Worse, the scan time drifts a little each day, because the next slot is measured from the last completed run rather than anchored to a fixed hour. Set a start and stop hour to hold the schedule still.
Random jitter makes the source scan slightly later than the chosen time, never earlier, so several sources do not all fire at once. A large jitter can push publishing outside the window you intended — the system warns when it is big enough to do that.
Which articles and how many
| Priority | What it does | Best for |
|---|---|---|
| Newest | Sorted by publication time, taken from the top. | Current affairs. |
| Random | Picks arbitrarily among the new articles. | Running several channels without them producing identical videos. |
| Oldest first | Whatever arrived first gets rendered first. | Making sure nothing is skipped. |
| Take everything | Every new article is rendered; only the daily cap holds it back. | Low-volume sources you want fully covered. |
Both numbers count ARTICLES, not videos
Articles per scan and maximum articles per day both count articles. One article can become several videos if you selected several templates, so the real video count is higher.
Setting articles per scan to 0 means take everything new in that run. Once the daily cap is hit, further articles are dropped and not carried over to the next day.
Filtering
- Only articles published within N hours — skips older ones. Blank means no age limit.
- Only articles containing keywords — blank means take everything.
- Exclude articles containing keywords — evaluated after the list above, so exclusions always win.
The preview marks each article as Taken or Dropped, so you can verify the filters before going live.
What happens once a video is rendered
| Mode | Result |
|---|---|
| Publish immediately | Goes to the channel as soon as the video finishes. |
| Wait, then publish | Publishes after a fixed delay in minutes from when the video finished. |
| Keep as draft | Saves to the library for you to publish by hand. Use it while you still want to review. |
In immediate mode, spacing between channels staggers each account's post by a few minutes so they do not all appear at once and look automated.
Credit safeguards
No daily cap is the biggest risk here
A high-volume source, running around the clock, with no daily cap, multiplied by several templates, can burn a great deal of credit in a single day without you noticing. The system warns when you leave the daily cap unlimited.
Stop the source when the wallet falls below is the last line of defence: set a threshold and the source halts once the balance drops under it. Zero means no automatic stop.
Start conservatively: one source, one template, three articles a day. Run it for a few days before opening it up. See also how credit is calculated.