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

# Build a Sales Research Agent with AIsa

> Plan a read-first company, people, contact, creator, and outreach-research workflow with clear consent and action boundaries.

A sales research agent should begin with read-only discovery and enrichment. Sending messages, adding contacts to sequences, or changing a CRM are separate authenticated actions with external side effects.

## Choose the capability

| Need                                    | Start with                                                          |
| --------------------------------------- | ------------------------------------------------------------------- |
| Company, people, and contact APIs       | [API Reference](/docs/api-reference)                                     |
| SEO and competitor opportunity research | [SEO Keyword Research](/docs/agent-skills/seo-keyword-research)          |
| Creator and contact discovery           | Relevant WaveInflu API pages in the [API Reference](/docs/api-reference) |
| Broader company or market evidence      | [Research agent use case](/docs/use-cases/research-agent)                |
| Task-oriented reusable workflow         | [Agent Skills](/docs/agent-skills)                                       |

Use direct APIs when the application must control filters, scoring, deduplication, or CRM mapping. Use a Skill when the process is reusable and the agent should follow a consistent research method.

## Recommended workflow

### 1. Define the target profile

Specify the organization, role, geography, industry, company size, or creator profile that qualifies. Avoid collecting unrelated personal data.

### 2. Discover organizations or people

Start with broad search criteria, then narrow the result set. Keep source identifiers and URLs so records can be traced and deduplicated.

### 3. Enrich only shortlisted records

Enrichment may be billable and may return personal or professional contact information. Apply data-minimization rules:

* Request only fields needed for the workflow.
* Avoid enriching every discovery result by default.
* Record the source and retrieval time.
* Respect applicable privacy, consent, and outreach requirements.

### 4. Score with explicit criteria

Keep qualification logic inspectable. Separate:

* Source data.
* Derived scores.
* Model-generated interpretation.
* Missing or uncertain fields.

Do not let a model invent revenue, headcount, job title, contact details, or buying intent.

### 5. Prepare an outreach brief

A research result may include company context, relevant triggers, suggested messaging, and source links. Drafting is not the same as sending.

### 6. Gate external actions

Before creating a contact, adding a person to a sequence, sending a message, or changing an external system:

1. Verify the connected account and required permission.
2. Show the intended action and target.
3. Obtain confirmation where policy requires it.
4. Execute the narrowest operation once.
5. Verify the external result.

Read access does not imply write permission.

## Minimal documentation set

* [Capabilities by Goal](/docs/by-goal)
* [Capabilities by Interface](/docs/by-interface)
* The exact company, people, enrichment, or creator API pages
* [Authentication](/docs/guides/authentication)
* [Security](/docs/guides/security)

## When AIsa may not be necessary

A direct CRM or data-provider integration may be simpler when the team uses one provider and already has stable enrichment, consent, and outreach processes.