Why does the interface language on my profile matter?
Because it isn't decorative. Whatever you set as your account language becomes the default your agents speak back in — chat replies, generated reports, drafted site copy, all of it. Flip your profile from English to Spanish and the next agent conversation you open just... speaks Spanish, no toggle required. Want one specific build to run in a different language than your account default? That's a per-project override, not a profile setting. The full mechanics — including how agents can detect language from your prompts even when you've touched nothing here — are in One platform, twenty languages.
Can I use both Google sign-in and a password?
Yes, and it's additive, not a swap. If you signed up with Google and later set a password "just in case," you can still sign in either way afterward. There's no path where adding a second method locks out the first.
Why are credentials split across three separate pages instead of one list?
Because a single junk-drawer "Credentials" page is exactly how you end up granting a build agent access to something it has no business touching, or hunting for a deploy password in forty rows sorted by nothing in particular. So each credential type lives only where it's actually used:
| Store | What it holds | Why it's separate |
|---|---|---|
| Google Accounts | Service accounts powering analytics connections — GSC and GA4, tied to specific domains | One account can cover both GSC and GA4 for a property; you don't need two connections per domain, just the right scopes granted on Google's side. A domain stuck on "pending" here is almost always a Google Cloud permissions gap, not a broken flow. |
| Deploy | SFTP targets — host, port, credentials — for your servers | One is read access for reporting, the other is write access to a live server. Mix them and you'd scroll past four service accounts just to rotate one SFTP password. |
| Store & social | No settings page at all | Store credentials show up in the ship flow when you're actually pushing a product live; social accounts show up in the marketing surface when you're setting up a posting schedule. |
Meaningless outside their one use case, so they surface in context instead of living on a page you'd visit twice a year.
Are my credentials actually isolated from other tenants?
Every credential in all three stores is scoped to your tenant, used only for the purpose it was added for, and deletable in place — click delete, it's gone, not soft-disabled. If you want the actual guarantees behind that — infrastructure-level isolation, what an agent can and can't reach — read tenant isolation. Worth it before you put anything with real stakes into these pages.
What happens to my billing if I upgrade mid-cycle?
It applies immediately, and any unused value on your old plan credits against the new one. Eight days into a monthly cycle and you upgrade? You're not double-paying for those eight days — the system nets it out. That's what makes upgrading low-stakes: no "wait for renewal" tax for wanting more capacity right now.
What about downgrading or canceling?
Both apply at the next cycle boundary, not instantly, and that's deliberate. You keep your current plan's allowance and features through the period you already paid for. We could've built instant downgrade with a partial refund, but that's worse for almost everyone — you'd lose capacity mid-build, potentially mid-run, for a refund that's usually a few dollars. Cancel on day 3 of a 30-day cycle and you still have the other 27 days.
Where do I actually learn how credits get spent?
Not here. How the monthly allowance resets, why credit packs never expire while the allowance does, the draw order between allowance and packs, and what happens if a build is mid-run when your balance hits zero — that's real substance and I'm not going to compress it into a paragraph and get it wrong. It's the whole subject of Credits, explained plainly. Read it before you're surprised by a mid-build stop, especially if you're running more than one domain.
Does the AI media page control every image an agent generates for me?
It sets your account-wide defaults — provider, quality tier, style preferences (photographic, illustrated, a specific palette). It lives at the account level rather than being reconfigured fresh every chat because consistency is usually what you want: five builds for five sections of a site should probably share a visual baseline unless you've got a reason otherwise. When you don't, per-chat overrides handle it — a seasonal campaign page that shouldn't match your evergreen brand imagery can override the default without touching your global settings. The override lives with the chat, so it doesn't leak into your next project by accident. Full mechanics in the image chapter.
