The AI campaign generator for agencies, built into the Main Library

Describe the campaign in up to 600 characters: a product, an offer, an audience or a moment to build the themes around. The brief is optional, but one specific sentence beats leaving the model to guess. Then pick an angle (Core positioning, Promo & offer, New audience, Objection & proof or Seasonal moment) and what each theme gets. Keywords can be switched off. Copy comes in five packs: Google Search ads, Meta ads, Email, SMS and Social posts. Creatives run from none to three per theme. Choose everything and the estimate reads about 200 copy lines and 15 creatives, so you can trim before anything is drafted.

The draft is grounded in the client rather than in a generic prompt. For anyone whose role reaches client profiles, it reads the Profile and the latest completed company research, and the card says which research it will use, or warns when there is none or a run is still going. A figure in a draft, such as a percentage, a price or a rating, has to appear in the brief or the client's verified record, and drafts that break the house writing rules are dropped before they land. Without profile access the AI works from your brief and the client's existing Library, and the card says so.

Drafts are written straight into the ordinary Library tools: themes arrive with the status Draft, keyword lists and creatives carry an AI campaign tag, and copy lines are filed under their theme. None of it is shared with the client, and nothing is sent, published or launched. Anyone who can see that client's Library can see the drafts, which the start panel says plainly, so treat a campaign like a teammate's work in progress.

Review follows the same path as work written by a person. Open each tool, edit or delete what is off, and submit what is ready for internal approval; an asset needs an approval, or an explicit skip by someone with approval rights, before it can be shared. A campaign's drafts have no review record until someone submits them, so the Approvals inbox stays quiet until your team acts. Portal access is part of the Scale and Enterprise plans, and where a client has one, an approved asset can be shared into it and sent for the client's own answer. If a campaign stops partway, what landed stays and the card names what is missing.

Pick a client and the page opens with its Library band: themes, copy lines, keyword terms, creatives and the reviews waiting, beside how many assets are shared in the client portal. Under the AI Campaign card, Library at a glance lists twelve rows, one for each tool from Themes to Approvals, with a sentence on what exists in it, such as how many target and negative lists hold how many terms, or how many emails are still drafts. Each row has an Open link to its tool, so a gap turns into an action.

Each number follows the definition the tool's own switcher uses, so the two agree; where a tool holds more than a thousand rows for one client, the overview is the exact count. A tool your role or plan withholds shows a lock, which means hidden and not empty. A read that failed shows a dash and never a zero. That distinction matters when a quiet row could mean no keyword lists or no access to see them.

Choose All Clients in the switcher and the page becomes a portfolio: a headline strip summed across the roster, then a table with a row per client and a column for each Library family, plus company research where your role reaches it. Sort by client or by any family to find the account with no creatives, the client with the most reviews waiting or the one that has never had a keyword list. Select a name and the page switches to that client's breakdown.

Totals say how much they cover. If your role reaches a family for only some clients, the headline reads across 3 of 12 clients, for example, instead of passing a partial sum off as the whole, and a withheld cell shows a lock rather than a zero. Below 1,024 pixels the table stacks into cards. For an agency lead it is the Monday view: who is thin, who is waiting on a review and where the next campaign should go.

Every Library tool files its work under a theme, so the By theme table reads as a coverage map. Each row is one angle with the keyword lists, copy lines, creatives, blog posts, script sets, landing pages, forms and emails filed under it. A theme with copy but no creatives, or creatives but no keyword list, shows without opening a single tool, and each theme row links to Themes for editing.

The table lists up to 20 themes per client and says when it has cut the list. Archived themes fold into a group of their own, kept as history, and work with no theme sits under Unassigned, so a deleted angle never hides an asset. When an AI campaign finishes, its new draft themes join the table with what was drafted under each, unless the client already has more than 20 current themes, in which case the table lists running ones first and says how many it did not show.

The Company research row at the foot of a client's page shows the newest run as a chip: Not started, Queued, Running, Publishing, Complete, Partial or Failed. Beside it sit how the run began, at onboarding or by hand, and a relative time such as completed 3d ago. A link opens the client's Research tab for the full record. Because campaigns draw on the latest completed research, this is the line to check before you press Create AI campaign.

Research status is client-profile data, so it follows profile access. Roles without it see no research row, no research column in the portfolio and no run named on the AI Campaign card. The row shows status and timing only, never the findings themselves. That keeps the page safe to open across a team where only some people handle the client relationship.