Agent teams & Discovery
Find something worth building, then keep it improving after it is live.
Discovery — deciding what to build
Open Workload → Discovery and describe a direction in your own words: "problems freelancers complain about", "growing businesses in Dubai". Add a region if it matters. Leaving it empty is fine — you get an open-ended search.
Five research agents then work in sequence across trends, social chatter, pain points and the wider web, and a final one turns the findings into briefs. It takes roughly 10 to 25 minutes and you can leave the page.
What a brief gives you
You get up to six Opportunity Briefs. Each names the problem, rates how confident the research is, lists the demand signals behind it and the gaps in what competitors offer, sketches a first version worth building, and links the evidence it relied on so you can check it yourself.
From a brief you can build it straight away, validate demand first with a waitlist page that measures real interest, draft a paid campaign, draft social posts, or dismiss it.
Agents that work on what you shipped
Once something is live and connected to Search Console and Analytics, the SEO team works on it:
| Agent | What it produces |
|---|---|
| Analyst | What is underperforming, read from your own search and analytics data |
| Research | Keyword opportunities worth going after |
| Strategy | A content plan built from those opportunities |
| Content | Drafted copy and metadata, staged for you to review |
Tasks, runs and schedules
A task is a saved recipe: a name, which of your domains it covers, which agents take part, and optionally when it should run by itself. A run is one execution of that recipe, which you can follow stage by stage, stop, or delete afterwards.
Schedules gathers everything that fires automatically into one table, with an enable switch and when each last ran and next runs, shown in your own timezone.
Teaching the agents
- Agents — give an agent standing instructions, attach reference files, or restrict it to certain domains. Whatever you set applies every time it runs.
- Skills — instruction blocks you write once and attach to whichever agents should follow them.
- Knowledge — your own reference documents, which agents can read when you attach them.
- Learnings — what the agents concluded from past runs, newest first, so you can see why later runs behave differently.