Three months ago I opened x-rush.ai with twenty minutes and a deadline: a tarot pull and a meme for my newsletter's morning thread. I burned four of my five opening-day credits just mapping the place — four tabs, thirteen features, no sign that said "start here." Both pieces shipped. This is the path I hand beginners: real clicks, real inputs, and the price of every run before you commit a credit.
Step 1: Create an account
Head to /en/auth/register. Username, email, a code from your inbox, password. No credit card anywhere in the form.
Free accounts get 5 credits a day. More useful: every feature shows its fixed price before you confirm a run, and a run that fails or gets blocked releases the hold and costs nothing. The number sits on the confirm button.
Step 2: Pick your lane
The four tabs, mapped:
- Text: HTML Studio for pages, AI Chat for characters, Automations for scheduled jobs
- Image: Image Studio, Character & Scene, Meme Image, Design Assets
- Video: Video Studio, Meme Video, AI Drama
- Audio: Voice Studio, Music Gen, Sound Design
Start with Text → AI Chat. Shortest path from typing words to something that feels like a trick.
Step 3: Your first reading
AI Chat opens on a Character list: a Tarot Reader, a Trivia Master, a Fitness Coach. I picked the Tarot Reader, since somebody had to.
Type: "What does my day look like?"
A full reading streams back — card, orientation, advice — in the character's own voice. There's no model picker in this flow; a routing layer sits behind every reply, deliberately invisible, and the same question can ride a different engine next week without your workflow changing.
One thing confused me. I asked the Trivia Master something, went back to the Tarot Reader, and found yesterday's reading exactly where I'd left it. State leak, I assumed. Wrong way around: every Character owns one continuous chat history, and switching resets nothing. Deliberate. Not a bug.
Step 4: Generate a Meme Image
Image tab → Meme Image. I typed "When the deploy passes CI and production falls over anyway," picked a two-panel template, and hit generate. One image comes back for 1 credit; each regenerate adds another candidate to the same batch at 1 credit each, and you click whichever you want to download.
Start from a reference photo, a template, or nothing at all; the side-panel copilot helps find the joke if you don't have one. I've made worse memes in Photoshop in an hour. This runs at 1 credit.
Step 5: Where your stuff lives
Every confirmed run saves into your Project, and the Project page aggregates all creations across every feature in one view. Uploads join your account media library only when you tick the box — a 500 MB pool shared across all projects, separate from generated results.
Step 6: Sharing, and what you actually get
Everything stays private until you publish it. And older guides promising a share button that mints a short link for any result: that's gone. The current model:
- Media results (images, video, audio, memes): download the file, or publish to that feature's Community/Explore and share the entry's detail page. No standalone URLs for these.
- HTML pages: the exception, each published page getting its own stable public address that opens for anyone, no login.
- AI Chat: pick specific messages and mint a read-only share link.
Downloads come down clean on every plan; no watermark on your own files.
What things cost
Chat isn't metered per reply: successful assists stack across your whole account — chat, copilots, prompt tune-ups, all of it — and every 10th one costs 0.5 credits. A failed reply bills nothing. Everything else is fixed price per confirmed run:
| Generation unit | Credits | |---|---| | Text | 0.5 | | Image | 1 | | Ordinary audio (voice, sound design) | 2 | | Music | 5 | | Video | 10 |
Failed, blocked, or undelivered runs release the hold; re-running something you didn't like is a new paid run, so read the candidates before you burn a reroll.
Plans cap simultaneous runs: 1 on Free, 2 on Lite, 3 on Pro.
What's next
Once the basics sit comfortably: Image Studio for photo retouching and restyling, AI Drama for script-to-video, Voice Studio for recording a sample and making it say anything in your voice.
One sleeper: Automations. Bind a Character to a scheduled job (mine: a morning tarot pull). Each run is a 0.5-credit text generation delivered into that Character's chat history with an in-site notification. No email, no push, by design.
The daily 5 credits cover a real learning curve. When you'd rather queue work than wait on it, Lite is $15 and Pro starts at $29.
Account to first creation in one sitting. Go make the thing you opened four tabs for.