Priced by volume. Never by features.

Start on any plan and the first fourteen days are free. We ask for a card and authorize $1 to confirm it’s real. That hold is never charged and disappears within a few days. We email a reminder seven days before the trial ends; cancel before day fourteen and you pay nothing. Otherwise the plan starts and bills monthly.

One credit is one SERP research run: the agents read your keyword’s page-one results live and turn them into charted evidence and a ranking-ready brief. You’re charged only when the run succeeds. A failed run refunds its credit automatically. Refreshing existing research uses a credit like a new run.

No. Credits reset with each billing cycle rather than accumulating. If you regularly run out, the next plan up is usually cheaper than the time spent rationing runs.

Each plan includes a set amount of AI usage for the writing side (drafting, edits, agent runs, image generation), measured in plain dollars: $6 a cycle on Pro, $20 a cycle on Startup, $80 a cycle on Business. You’ll see a warning at 80% of it. At the cap you choose: add your own API keys, upgrade, or wait for the cycle to reset. Nothing ever bills past your plan price.

Add your own OpenAI, Anthropic, Google, or DeepSeek API keys and the calls made with them stop drawing your allowance. The provider bills you directly, at their price. We add nothing on top. It works per provider and per call, not as a mode: a provider you’ve keyed runs outside the allowance, everything else keeps spending it. Keys are validated when you add them and stored encrypted.

Yes. Keeping the two apart is the whole point. Your usage page splits every day’s spend into what came out of your allowance and what went to your keys, breaks it down by provider and by feature, and keeps a month-by-month history. The keys page adds a per-provider chart, month-to-date totals, a per-model table with call counts, and a CSV export. One honest caveat: those dollar figures are our estimate, priced from our own per-model rates. Your provider’s invoice is the real bill.

Pull a key out whenever you like. The next call for that provider simply goes back to your included allowance. And if a key ever stops authenticating, we mark it invalid and fall back to the allowance rather than failing your draft mid-run.

No. Every plan runs the same product: the full research pipeline, unlimited documents, the whole toolkit, and all eleven models. Plans differ in exactly three numbers: price, research credits, and included AI usage.

Yes. Upgrades apply immediately. The bigger allowance is available at once and the price difference is prorated. Downgrades take effect when the current billing cycle ends. All of it happens in the app, no email required.

Really. Documents, projects, briefs, versions: none of them are counted. The AI work inside a document draws on the allowance (or your keys); the documents themselves are unlimited.

No. Instead of a limited free tier, the trial gives you the whole product for fourteen days (real research runs, real drafts), so you decide on evidence, not on a demo.

Visa, Mastercard, and American Express, billed monthly in USD through Stripe.

Cancel from the app and nothing changes that minute: the plan runs to the end of the period you’ve already paid for, and you can reactivate any time before then. After that it ends: no further charges, and no refund for the stretch you didn’t use. Cancelling doesn’t wipe your work: your documents, research and briefs stay in your account, and the app narrows to billing, settings and support until you subscribe again. Content nobody touches for 90 days may be cleared out after that. (Cancel during the trial and it ends right away, with nothing charged.)

Cancelling a plan and closing an account are different things. Ask from your account settings and we delete the lot (documents, research, briefs, keys) outright. That one is immediate and can’t be undone, so export anything you want to keep while your plan is still active.

Not by nine of the eleven, and we’ll name the exception rather than bury it. WordCrafter calls the providers’ paid APIs, and OpenAI, Anthropic, and Google don’t train on data submitted through theirs. DeepSeek’s API terms are the exception, so if that matters for your content, pick any of the other nine models. Your documents and your research are yours either way: we never train on them.