Before you start
Step by step
1
Find the exact values the agent matches on
Industries must be exact strings, so resolve them first:Round types come from a fixed list:
pre_seed, seed, series_a, series_b, series_c and so on through series_j, plus private_equity, debt_financing, grant and others listed in the API reference. Countries are full English names such as "United Kingdom", and headcount uses exact bands such as 11-50 and 51-200. A value outside these lists is still accepted with 201, and the agent then quietly finds nothing. See Finding accepted values.2
Create and start the funding agent
industries, companyCountries or companyHeadcount is required, and each takes several values. start: true creates the agent and starts it in one call, which also reports any configuration problem straight away. To review a draft first, leave start out and activate it later, as in Job change outreach.3
Find the agent's company list
GET /business/agents returns the agent’s results list:contactsFound counts the companies delivered so far.4
Read the funded companies' domains
total is how many company rows it has. Once the list grows past 1,000, page through it with start (1, 1001, 2001 and so on). The agent keeps adding companies, so keep a record of the domains you have already searched and only search the new ones each time. GET /business/lists/{listId}/rows returns each company’s name, industry, headcount and country if you want to filter further. The agent records the company itself; the round, amount and investors are not written to the list.5
Find the people you sell to at those companies
Pass the domains as Then save them with work emails into one list that every run appends to:Poll
job_company_website, up to 1,000 per search, with the roles and seniority you sell to. Preview the count first:GET /business/jobs/{jobId} with the jobId from the 202 until status is completed, then read metrics on GET /business/lists/{listId}. Very new or very small companies may not be in the people database yet. A Deep Research run can find their leaders instead.6
Point a dynamic campaign at the people list
Do this once, after the first save has filled the list:Wait for initialization, preview and approve as in steps 7 to 9 of Build an AI outbound campaign from a prompt. The agent records the company, not the details of the round. To name the round and lead investor in the email, write that line with a smart column and merge it in with a column token. Add each day’s people in batches, as described in Personalize every email with AI research, so nobody is enrolled before their line exists.
Run it every day
The agent delivers on its own schedule. Once a day, repeat steps 4 and 5 for domains you have not searched yet. Saving with the samelistName appends to the people list, and the dynamic campaign enrolls the new contacts within about 30 minutes, up to 200 at a time.
What it costs
Troubleshooting
Run the whole workflow
Python script: daily funding sync
Python script: daily funding sync
Do it from Claude
With the Fuse MCP server connected:Set up a signal agent for software companies in the US and UK with 11 to 200 employees that announce a Series A or Series B. Then find the finance and operations leaders at the companies it has delivered so far, save them with work emails to “Funded - buying committee”, and draft an AI campaign on that list that enrolls new people automatically. Don’t launch it.The assistant sets up the signal agent and reads the companies it has delivered, then searches for the people, saves them and drafts the campaign. Creating a campaign never launches it.
Related
Create an agent
Every agent kind and its configuration fields.
Build a list from search
People search filters and the accepted vocabularies.
Job change outreach
The same always-on pattern with an agent that delivers people.
Website visitor outreach
The same domain-to-people step, driven by who visits your site.