AI lead search: turn natural language into B2B filters

1. Use AI to build the search, not manufacture the data

Natural-language lead search is useful when you know the audience but do not yet know which filter fields express it. Describe the buyer role, company type, market and required contact channel in one sentence. Feediqo converts that request into the same structured filters available in the manual builder.

The model does not generate a list of imagined people, email addresses or companies. It selects supported filter values, and the directory search returns the matching records. That distinction keeps the targeting visible and lets the user correct an interpretation before using the results.

2. Write a short audience description with observable criteria

A useful prompt names the role, market and company context that matter to the offer. “Owners of construction companies in Texas with email” is more actionable than “good prospects for my agency” because each phrase can map to a field the directory can search.

Avoid asking AI to infer private intent, budget or dissatisfaction. A database may contain a role, industry, location or technology signal, but those fields do not prove that a prospect is ready to buy. Keep the prompt tied to facts you can review.

• Role or seniority: owner, founder, director or another relevant level.
• Company context: industry, company name, website or business type.
• Market: country, city or region when geography matters.
• Contact route: email, phone or LinkedIn only when required.
• Exclusions: remove criteria that are assumptions rather than recorded fields.

3. Review every generated filter before accepting the audience

The generated filters remain editable. Check whether a phrase was interpreted as a title, seniority, company, industry or location. A city and a state can share names, and job-title language can be broader or narrower than intended, so the visible filter chips are an important review step.

Run a small result sample and inspect several profiles. If the matches are too broad, add one criterion that describes the market more precisely. If the result is too narrow, remove the least important condition rather than replacing the entire prompt.

4. Qualify the returned records before revealing contacts

A filter match means the available record satisfies the search logic; it does not confirm buying intent. Review the current role, company, location and reason the prospect fits before spending a reveal credit or saving the record to an outreach list.

Keep useful prospects in a saved list with notes that explain the fit. Remove duplicates, existing customers, competitors and suppressed contacts before the list enters a campaign. The AI search step makes audience building faster, but qualification still belongs to the user.

5. Move a reviewed people list into a connected campaign

After the audience is reviewed, reveal suitable contacts and save the selected people to a list. From there, connect one or more Gmail inboxes in the Campaigns workspace, choose the sender and reuse an approved message template.

Feediqo can schedule up to three automatic follow-ups and lets the user pause or resume the sequence. It does not read replies, so monitor the connected inbox and pause the campaign when someone responds. Search automation and sending automation remain separate reviewable steps.