AI Assistance / Agents
Workers that run on a schedule, touch only your data, and report to a person.
An agent here is a standing job: a watcher that notices, a report that writes itself, or one you compose from your own data blocks. Each has an interval, a monthly budget, a scope, and a run history with the cost of every run.
Demo workspace. Invented brand and figures.
Four that notice, two that write.
Search rankings watcher
about once a day
Google ranks of your tracked keywords. Flags page-one entries and exits, drops of three places or more, new, and lost keywords. Ignores one-place wobbles.
AI visibility watcher
about every two days
What the assistants say about you against the previous pass: mentions, answers, share of voice, and the citations you gained or lost.
Ads campaigns watcher
about once a day
The live Google Ads mirror. Campaigns that stopped serving, status flips, budget changes, spend jumps, with the real numbers.
Market monitor
about every two days
Competitor moves on your keywords and clusters, read from the stored search-results reports and ranking history, written for a marketing director.
Weekly operating review
every week, a document
Reads across rankings, ads, content, social, pipeline, and spend. Writes what changed, why it likely changed, and what to do next week.
Executive report
every month, a document
The period’s narrative for the board: what moved, why it matters, what happens next. Runs on the reporting model you chose.
A finding is a paragraph: what moved, why it matters, and what to do. Under it, four buttons. Create the task, draft the response, acknowledge, dismiss. The severity is the agent’s call. The decision is yours.
What happens when an agent wakes.
Schedule
Due, or on demand
The scheduler wakes at six in the morning, universal time, and queues every agent whose interval has passed. Or press Run now.
Gather
It reads its sources
Only the data blocks it was given, only for the brand it runs on. The snapshot is compared with the last one.
Skip
Nothing changed, nothing charged
No new data or no movement means no model call at all. The run is recorded as unchanged, at no cost.
Write
Findings, or a document
Watchers post findings with a severity and a suggested action. Report agents write a document you read from their card.
Tell
Mail and ledger
After a scheduled run, the administrators get a summary: what it read, what it found, what it cost. The cost lands on the ledger.
Switched on, and on the record
The Organization tab shows what runs for everyone: each agent with its interval, its budget, and an Edit and test button. The Runs tab keeps every scheduled run, for every brand you can see: when, what it read, what came out, tokens, cost. Nobody has to ask what the agents did last night.
Compose your own from eight data blocks.
A custom agent is a brief plus the data blocks it may read. You are offered only the blocks your role can already see, and at run time the brand’s module has to be on. Share it with the organization or keep it personal.
It gets the same machinery as the standard six: a schedule, a budget, a scope, the zero-cost skip when nothing changed, and a line on the runs page.
The eight blocks. Pick any mix.
Six dials on every agent.
Four levels: platform, organization, brand, person. Each setting resolves from the closest level up, so you can override the brief and keep the organization’s model.
A monthly cap in dollars per agent, checked before the run, on top of your daily cap and the organization’s monthly cap. The tightest one wins.
Hours between runs. The catalog ships a sensible default per agent; change it per organization, per brand, or for yourself.
Rewrite what the agent is told to look for. Pick the model per agent, or inherit the module default.
Run it against one brand’s live data and read the output without posting anything to the feed.
Both lists start empty. Agents are added from the catalog. A member without the right asks with one click, and the administrator gets the mail.
Built, partial, not built
The catalog lists 96 agents across six families, each marked as built, partial, or not built, with a note that says why. The same catalog our customers see in the product.
The same data, one door over.
Nothing here is an integration. These run on the same account, the same brand, the same ledger, so a finding on one page becomes work on the next.
Not built yet, said first
- Findings appear in context on two module pages today, Ads and Market Monitor. Elsewhere, read them on the Runs page, the Watchers dashboard, or ask for them.
- No user-set alert thresholds and no scheduled distribution of reports by mail yet.
- Another 11 agents are offered tailor-made, on request. They are not product features.
The catalog lists every agent as built, partial, or not built. Checked against the product on September 3, 2026.
Usage is metered in tokens from a prepaid wallet, with no seat fee. The rates are on thepricing page.
Switch one on. Read what it found on Monday.
Add an agent from the catalog, give it a budget, and let it run. You’ll read the first findings the next morning, with the price next to them.
No subscription. No seat fee. No card needed to request an account.