The jazzy stuff

The headline features are easy to list. These are the quiet ones - the details that decide whether the output is usable or just a demo. Most tools skip them.

The text never overflows

Branded card text is measured against the real rendered font and shrunk until it fits - hundreds of renders checked, not one margin blown. No image model guessing at spelling, no headline cut off on Instagram. What you typed is what lands.

It fails out loud, never quietly

Hand it a dead link or an empty source and it says so, with the reason - it never invents content to look busy. A failed generation is marked failed, with the provider’s real reason. It will never serve you template copy dressed up as the real thing.

Twelve brands, no bleed

Voice is per brand. Run one client or twelve and none of them start sounding like each other on the same Tuesday. A model key carries no memory of you - only what we send applies.

Training is never lost

Every change to a brand’s voice snapshots automatically, with a plain-English diff, and any save-point is one click to restore. Nothing you teach it can be thrown away by accident.

The score shows its working

The Echo Score isn’t a bare number - it’s a number with the reason behind it (“strong hook, slightly abstract takeaway”). A number you can argue with, which is the point.

It picks the right shape for the platform

Open any image and it tells you its resolution, ratio, and which platforms that shape suits - and when you’re picking an image for a post, the ones that fit the platform you’re editing light up green. See Image sizes & placements.

The AI-tell filter

A universal pass that strips the words that read as machine-written before anything reaches you - on top of your own banned words.

Refine argues like a writer

Refine is a real back-and-forth, not a text box. Tell it what’s off, read the next pass, push back, accept when it’s right - and attach a reference image or doc mid-conversation and it reads that too.