Client settings - cadence, limits, tier & KPIs

A client's settings page is where you configure their account. It's grouped into four zones, calmest first, riskiest last.

Profile

The basics: name, website, contact, logo, country (for holiday-aware scheduling) and currency. This is also where you set the account tier - which controls feature access, including whether Creator's Studio is switched on.

Go-live date

An optional date marking when the client's site or tracking went live. Before this date, the Brain treats traffic, conversion and spend swings as pre-launch noise rather than real movements - so it won't flag launch artefacts (such as a "+1,798% sessions" jump on a freshly launched site) as things worth acting on. Leave it blank if the client was already established when you connected them.

Setting or changing the date also re-reads past movements, so earlier swings are re-classified against the new date straight away.

Configuration

• Data refresh cadence - manual, weekly or daily. See Data refresh & cadence.
• Monthly usage limits - per-feature caps (reports, copywriter, social, market research and more), shown against current usage. Leave a field blank to inherit the platform default.

Advanced

Brain operations for the account:

• Backfill - seed the Brain from existing platform data and documents.
• Manual reflection - force a reflection run.
• A link to the client's Brain page for a deeper look.

See Operating & monitoring the Brain.

Connected AI

Whether this client's team may connect their own AI tools - Claude, ChatGPT, Cursor - to read their workspace, and how far those tools may reach. On by default, for every client: connecting their own AI to their own data is part of the product. Switch it off for a client you'd rather keep closed, and use the cap and the review switch to shape what connected tools may do.

• Let this client connect AI tools - the master switch, on unless you turn it off. While it's on, an AI tools section sits on the Connections tab of their own dashboard, beside the platforms they already connect, where they can sign in from claude.ai or ChatGPT or create a token. Turning it off removes that section; your own connections to this client keep working either way, because those are governed by who holds them rather than by this switch.
• Maximum permission level - the widest level they may hold: Read-only, Read + ask (adds questions to the Brain, drawing on their AI allowance), or Collaborate (their AI can also save knowledge and suggest corrections). Leave it uncapped to let them choose. Lowering it applies immediately, including to connections that already exist.
• Review everything - nothing their AI proposes applies on its own: confident facts, logged moments and corrections all queue for review instead. Insights pulled from meeting notes always queue regardless.

The same card lists every live connection with who connected it, when it was last used and how many calls it has made, and lets you revoke any of them - for offboarding, a suspected leak, or a client who has asked you to pull it. Underneath, what their AI asks for most over the last 30 days is a genuine read on what they need: a client whose AI keeps reaching for one channel's data is telling you something worth raising in the next review call.

Everything on this card is about the client's own people. Your team's AI connection is set up once for every client at Connect your own AI to every client, not here.

Self-serve workspaces manage their own connections and are always allowed to connect; the cap, the review switch and revoke still apply to them.

The client-facing side is documented in What Connected AI is & how it works.

Danger zone

Destructive actions, admin-only and behind confirmation - most importantly deleting the client (which removes all of their data).

KPIs

Choosing which KPIs the client cares about lives on the Manage page (Goals & work zone) - those choices drive the client's dashboard focus and report emphasis.

Roles

Profile, Configuration and Advanced are open to your team; the tier selector and the Danger zone are admin-only.