The Amplicast Workflow: From Raw Content to Live Post

The Amplicast workflow has seven steps: connect a source, sync, generate or import, transform, review, publish, measure. That's the whole loop. Here's what each step actually looks like in the app, and what's running under the hood while you work.

Step 1 — Connect a source

A source is where your raw material lives before Amplicast touches it. Supported types include Notion, RSS, GitHub, a generic REST API, Supabase, a web scraper, and a CDN manifest. The point is that you point Amplicast at the place your content already exists, rather than pasting it in by hand.

In the app: Sources → Add source → pick the type → paste the URL or credentials → save. That's the full setup. Behind the scenes, the credential is encrypted at rest with AES-256-GCM before it hits the database. The sync worker validates connectivity on its next cycle, so you know immediately if something is misconfigured.

Step 2 — Sync

You do nothing here. Sync runs on a schedule you configure per source. You can also trigger it manually if you want to pull something right now.

Behind the scenes, the sync worker fetches new items with a per-source timeout, de-duplicates against everything already imported, and records any fetch error against the source directly. Nothing is silently dropped. If a sync fails, you see it.

Step 3 — Generate or import content

Two paths. The first is Autopilot: define a topic and a cadence (say, weekly), and Amplicast drafts a content package on schedule using a Claude model tuned with your brand voice and writing samples. You can also hit Generate now to skip the wait. If the AI provider is momentarily overloaded, generation retries with backoff and falls back to a secondary model — so a slow API doesn't kill your run.

The second path is manual import: Content library → pick a synced item → it becomes a draft package. Useful when you want to take a specific article or update and turn it into platform posts without waiting for Autopilot.

Either way, what you end up with is a package: a title plus a body that serves as the source material for the next step.

Step 4 — Transform for each platform

This is where one piece of content becomes several platform-specific drafts. Open the package, choose your target channels, and optionally toggle text transformation, image filters, crops, logo overlay, or AI-generated imagery.

The transform pipeline rewrites the caption to match each platform's character limits and tone, adapts or generates imagery, and produces one draft post per channel. LinkedIn gets a different caption than Instagram. X gets something that fits 280 characters. You're not doing that rewriting manually.

Media downloads are size- and time-bounded during this step. A malformed or oversized asset can't hang the pipeline indefinitely.

Step 5 — Review (optional)

If your project has a review step enabled, posts land in a pending review queue before anything goes out. An editor can approve, edit, or reject each post from that queue. Review state is enforced per organization, so a reviewer only ever sees their own tenant's posts.

If you're a solo creator who doesn't need an approval step, you skip this entirely. It's a per-project setting, not something forced on every account.

Step 6 — Publish

Publish now or schedule for later. Scheduled times are stored as UTC-aware timestamps, so timezone errors don't silently shift your post by hours.

The publishing worker resolves the exact connected account by its stable internal ID, not by platform name. That means a post created for your LinkedIn company page can only ever go to that page, not a personal profile or a different account you've connected. On a transient platform error, the worker retries with backoff, honoring the platform's Retry-After header.

Every publish is logged with the platform's response — success or failure. Credits are deducted exactly once, only after the post is durably recorded as published. Platforms supported include Instagram Business (via the Graph API), Facebook Pages, LinkedIn personal profiles and company pages (tokens auto-refresh), X, TikTok, YouTube, Bluesky, and Reddit.

Tokens are encrypted at rest and refreshed automatically before expiry. Bearer and session tokens are stored only as hashes. If a publish fails, the account owner is notified and the failure is logged — nothing disappears quietly.

Step 7 — Metrics

After a post goes live, a metrics worker periodically pulls engagement data — likes, comments, shares, views — from the platform APIs and stores it against the post. You see it directly on the post in Amplicast. No switching between platform dashboards to piece together how something performed.

What the full loop looks like

Source pulls content. Sync keeps it current. Generation or import turns it into a package. Transform adapts it for each platform. Review (if you want it) catches anything before it goes out. Publish sends it, logs it, and retries if something goes wrong. Metrics close the loop.

Every step is scoped to a project and an organization. Nothing crosses between tenants. And if you ever want out, Settings → Delete account exports or permanently erases all your data, with a 30-day grace period to change your mind.

The workflow is designed so the actual writing is the hard part. Everything after that — the resizing, the reformatting, the reposting — runs without you.