> ## Documentation Index
> Fetch the complete documentation index at: https://www.landbase.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Build Audience

> Build targeted lists combining company and contact filters with semantic search, lookalikes, and AI qualification — all in a single operation.

<div id="breadcrumb-wrapper" />

<div
  style={{display: 'none'}}
  data-related-resources={JSON.stringify([
{ title: 'Qualify Leads', href: '/reference/tools/qualify-leads' },
{ title: 'Search by Meaning', href: '/reference/tools/search-by-meaning' },
{ title: 'Expand Lookalikes', href: '/reference/tools/expand-lookalikes' },
{ title: 'Tools overview', href: '/reference/tools' }
])}
  id="page-related-resources"
/>

# Build Audience

<div className="blog-post-content">
  Every other tool in Landbase's toolkit does one thing. Build Audience is where they all come together.

  This is the tool that takes your targeting criteria — who you're looking for, at what kinds of companies, matching what description — and produces a complete, qualified list of contacts. It's the output of a workflow compressed into a single step.

  ## What you can specify in one query

  **Who you're targeting (contacts):**

  * Job titles (with automatic expansion to similar titles)
  * Department and management level
  * Location and tenure
  * Skills and experience keywords

  **At what companies:**

  * Industry, size, revenue, geography
  * Tech stack and funding stage
  * Job postings ("companies hiring for data engineers")
  * Semantic description ("AI developer tools companies")

  **With smart expansion:**

  * Pin specific exemplar companies and find more like them
  * Provide seed accounts for full lookalike expansion
  * Describe your target market in natural language

  **With inline qualification:**

  * Define custom criteria ("Is this company likely B2B?", "Does this person manage a team?")
  * The system evaluates every record and writes verdicts into the results

  ## How it works

  Build Audience runs up to five processing stages in a single call:

  1. **Expand** — if you provided exemplar companies or seed accounts, the system expands to find similar companies and generates ranking boosts
  2. **Search** — all your company and contact filters execute against the search index, boosted by the expansion signals
  3. **Merge** — exemplar boosts and semantic boosts combine (strongest signal wins for each company)
  4. **Materialize** — results are assembled into a table you can view, sort, filter, and export
  5. **Qualify** — if you defined qualification criteria, AI evaluation runs and writes verdicts into the table

  You don't have to use all five stages. A simple firmographic search skips straight to stage 2. Adding exemplars activates stage 1. Adding qualification activates stage 5. The tool adapts to what you provide.

  <Expandable title="What's happening under the hood">
    The agent orchestrates multiple subsystems — search indexing, similarity expansion, embedding pipelines, and AI qualification — in a single function call. Exemplar domains are force-included and similarity-boosted via pre-computed expansion. Semantic phrases go through a multi-stage pipeline: embedding, nearest-neighbor seeds, then similarity graph expansion. All boost signals merge by taking the max score per company. If qualification is requested, a new table is created with AI qualification columns and optionally a preview qualification runs immediately.
  </Expandable>

  ## Why it matters

  Without Build Audience, you'd run each tool separately — search for companies, then find contacts, then qualify them, then export. Build Audience collapses that into a single request that chains the tools together automatically. The agent picks the right tools based on what you ask for.

  Every enhancement to the individual tools (better semantic search, deeper qualification, richer lookalike models) automatically improves Build Audience because it orchestrates all of them.

  ## Result limits

  Build Audience returns up to 10,000 records per query.
</div>