Starting a run

A research run takes a set of clients and an offering, then researches each client in the background. You can start a run from the Research runs page or from the New run shortcut.

Creating a run

1. Go to Research runs in the main navigation.
2. Click New run.
3. Give the run a name (optional). If you leave it blank, Client Path generates one from the offering name and date.
4. Select the offering you want to research against.
5. Add clients to the run using the client picker. You can:
   • Search by name or domain.
   • Filter by tag.
   • Select all clients that match a tag or filter.
6. Click Start run.

The run is created and queued immediately. Workers pick it up within seconds.

Picking clients

The client picker shows all clients in your workspace. You can narrow the list by:

• Searching by name or domain.
• Filtering by one or more tags.
• Filtering by whether the client has been researched before.

There is no hard limit on the number of clients per run, but very large batches (500+) may take several hours to complete.

What happens after you click "Start run"

1. The run record is created with status queued.
2. For each client, a client research record is created with status queued.
3. Worker processes pick up each client research and begin the pipeline: web search, document extraction, finding synthesis, scoring, report generation, outreach drafting.
4. As each client research completes, its status changes to completed (or failed if something went wrong).
5. Once all client research records reach a terminal state, the run status changes to completed and the rollup is generated.

Monitoring progress

The run detail page shows the status of each client in real time. The page refreshes automatically. You can see which clients are in progress, which have completed, and which (if any) have failed.

See also: research runs, reviewing results.