Locate the right workspace
Before any other capability can run, your assistant needs to know which workspace to look at. It can list the workspaces available to you and match a name you mention in conversation to the right one, so you can ask questions like “what’s happening in my Home Depot workspace” without ever handling an ID yourself. This is almost always the first step in a conversation and feeds into every capability below.Understand what a workspace is tracking
Once a workspace is in scope, your assistant can pull back the filters, sort options, and field metadata that workspace supports. This is what makes questions like “only look at news mentions from the last week” or “sort by reach” work, the assistant discovers what’s filterable before it runs the query, instead of guessing. You rarely ask for this directly. It runs in the background whenever you constrain a question.Get the shape of the conversation
Your assistant can summarize a workspace’s activity at a glance: volume over time, top tags, top authors and domains, the narratives driving the conversation, and where in the world it is happening. This is the right starting point for open-ended questions (“what’s going on with X this week?”) and for spotting spikes worth digging into. From here you’ll typically pivot into narratives or mentions to explain why something moved, or into geography to see where it moved.Explore narratives and their trajectory
Narratives are how PeakMetrics groups related mentions into coherent stories, with threat scoring layered on top. Your assistant can search across narratives in one or more workspaces, sort them by threat or momentum, and chart how a single narrative’s volume has moved over time. Common chains:- Analytics surface a spike → search narratives in that window → pull the mentions behind the top narrative to see what people are actually saying.
- Search narratives filtered by threat score → get the volume-over-time for the top one → decide whether it’s still accelerating.
Read the underlying mentions
Every summary and narrative eventually ties back to individual mentions. Your assistant can search mentions inside a workspace with the same filters and sort options the dashboard uses, page through results, and pull a randomized sample when you want a representative read of a large pool rather than the top-ranked slice. Use this whenever you need direct evidence, quotes, or links, analytics and narratives tell you what is happening, mentions tell you how it’s being said.Focus on a place
Your assistant can reason about where coverage is happening using the same normalized geography PeakMetrics resolves every mention to: region, subregion, country, state, county, and city. You can ask for a place by name (“Ohio”, “the Bay Area”, “West Africa”) and the assistant will resolve it to the right level before scoping any question, so you never have to think about codes or IDs. Once a place is in scope, geography composes with everything else:- Filter any question to a place. Narratives, mentions, top accounts, and top journalists can all be narrowed to one or more places at once, and a country selection automatically includes every state, county, and city inside it, so you don’t have to enumerate them.
- Break analytics down by geography. Workspace analytics can report volume by country, or drop to state, county, or city when you need finer resolution, to answer questions like “where is this story loudest?” without leaving the conversation.
- Locate specific mentions. Individual mentions carry the place they resolved to at each level, with coordinates for city-level results, so quotes and links can be grounded in a specific location when you cite them.
Identify who is driving the coverage
Two capabilities focus on people rather than content:- Top social accounts ranks the accounts producing the most mentions in a workspace, with reach and engagement signals, so you can spot amplifiers and coordinated activity.
- Top journalists does the same for news, surfacing the reporters shaping the story.
Apply your custom summary prompts
If your team has configured custom summary prompts on a workspace, your assistant can pull them and run summaries in your own voice and framing rather than a generic one. This is what makes MCP output usable directly in briefings and reports without a rewrite pass.Putting it together
The capabilities are designed to compose. A single request like “draft a threat briefing on election integrity narratives in the Midwest this week, and include the top three accounts amplifying them” naturally chains:- Locate the workspace by name.
- Resolve “the Midwest” to the right set of states.
- Search narratives in that geography, filtered to the last seven days and sorted by threat.
- Pull volume-over-time for the top narratives to describe trajectory.
- Pull mentions from those narratives for direct quotes.
- Rank top social accounts against the same window and geography.
- Apply your custom briefing prompt to shape the output.