<div class="article-tldr"><p class="article-tldr-label">TL;DR</p><p>You can score 1,000 leads against your ICP in about 20 minutes for roughly 3,000 API calls using Nqzai inside Clay. The article walks through the exact prompts and parsing steps to build that pipeline without a data team.</p></div>
<h1>How to Use Nqzai With Clay for AI Lead Generation</h1>
<p>Lead generation in 2024 demands more than static lists and manual enrichment. The combination of <strong>Clay</strong> (the go-to platform for multi-source data enrichment and outreach automation) with <strong>Nqzai</strong> (a specialized AI model for natural language querying and synthesis) creates a workflow that turns raw contact data into prioritized, action-ready leads—without requiring a data science team.</p>
<p>This article outlines a practical, step-by-step method for using Nqzai inside Clay to build a lead gen pipeline that is both scalable and precise.</p>
<h2>What You’re Building: The Core Workflow</h2>
<p>The goal is simple: feed Clay a list of prospects (names, companies, titles) and have it return a scored, enriched set of leads—with ranked fit scores, personalized messaging angles, and cleaned data—all processed through Nqzai.</p>
<p>Clay handles the orchestration, the API calls, and the data plumbing. Nqzai provides the reasoning layer: analyzing a prospect’s LinkedIn bio, company Crunchbase description, or job posting to assess fit against your ICP (Ideal Customer Profile).</p>
<h2>Prerequisites</h2>
<p>Before you begin, confirm you have:</p>
<ul>
<li><strong>A Clay account</strong> (any tier that supports HTTP requests and API integrations)</li>
<li><strong>An API key for Nqzai</strong> (available from the Nqzai platform; note that it uses a standard OpenAI-compatible endpoint)</li>
<li>A CSV or list of at least 10-20 test prospects (to avoid wasting credits on a broken workflow)</li>
</ul>
<h2>Step-by-Step: Setting Up the Nqzai Integration in Clay</h2>
<h3>1. Create a New Clay Table</h3>
<p>Start with a simple table containing the columns you want to enrich. At minimum:</p>
<ul>
<li><code>First Name</code></li>
<li><code>Last Name</code></li>
<li><code>Company Name</code></li>
<li><code>Company LinkedIn URL</code></li>
</ul>
<p>Upload your target list. Clay will auto-detect the columns.</p>
<h3>2. Add the Nqzai API Call</h3>
<p>Clay’s “Add Enrichment” menu includes a <strong>“Send HTTP Request”</strong> option. This is your gateway to Nqzai.</p>
<p>Configure the request as follows:</p>
<ul>
<li><strong>Method:</strong> POST</li>
<li><strong>URL:</strong> <code>https://api.nqzai.com/v1/chat/completions</code> (or the endpoint provided in your Nqzai dashboard)</li>
<li><strong>Headers:</strong></li>
<li><code>Authorization: Bearer YOUR_NQZAI_API_KEY</code></li>
<li><code>Content-Type: application/json</code></li>
<li><strong>Body (JSON):</strong> This is where the prompt lives.</li>
</ul>
<p>Example body for a lead-scoring prompt:</p>
<p><code>`</code>json { "model": "nqzai-pro", "messages": [ { "role": "system", "content": "You are a B2B lead scoring assistant. Evaluate the lead based on company industry, employee count, and job title. Score from 0-10 where 10 is a perfect fit for a SaaS company selling to HR tech." }, { "role": "user", "content": "Company: {{Company Name}}. Industry: {{Industry}}. Title: {{Job Title}}. Employees: {{Employees}}. Score this lead." } ], "temperature": 0.2, "max_tokens": 150 } <code>`</code></p>
<p><strong>Important:</strong> Use Clay’s “Merge Tags” (the curly-brace variables) to dynamically insert data from each row. This is what makes the workflow per-row.</p>
<h3>3. Parse the Response</h3>
<p>Nqzai returns a JSON object. You need to extract the score from it.</p>
<p>After the HTTP request step, <strong>add a “Parse JSON” step</strong> in Clay. Target the output field from the HTTP request (usually <code>result.data</code> or <code>output</code>). The actual content will be inside:</p>
<p><code>choices[0].message.content</code></p>
<p>Map that content to a new column, e.g., <code>Nqzai Score</code>.</p>
<p>If you want a numeric value, you may need a second parse step that extracts just the number (e.g., using a regular expression in Clay’s formula or a “Text Transform” step).</p>
<h3>4. Enrich with Additional Context</h3>
<p>A single score is useful, but Nqzai can do more. Create <strong>additional HTTP steps</strong> for:</p>
<ul>
<li><strong>Personalization:</strong> “Write a 2-sentence outreach email opener referencing this lead’s recent [Company News].”</li>
<li><strong>Intent signal extraction:</strong> “Does this job posting suggest they are hiring for a role that uses our product? Answer YES or NO.”</li>
</ul>
<p>Each step adds a new column of AI-processed intelligence.</p>
<h2>Three High-Impact Use Cases (With Nqzai)</h2>
<h3>Use Case 1: ICP Scoring at Scale</h3>
<p>Instead of manually reviewing 500 LinkedIn profiles, let Clay feed each profile’s <code>About Section</code> (pulled via a LinkedIn enrichment provider) to Nqzai. The prompt:</p>
<p><code>`</code> “Analyze this LinkedIn bio. Rate how likely this person is to purchase enterprise HR software (1-10). Consider job function, seniority, and company size. Output only the number.” <code>`</code></p>
<p>Result: a clean numeric column you can filter or sort.</p>
<h3>Use Case 2: Personalized Message Generation</h3>
<p>Standard token personalization (e.g., “Hi {{First Name}}”) is table stakes. Nqzai enables <strong>context-aware</strong> personalization.</p>
<p>Prompt:</p>
<p><code>`</code> “Based on the prospect’s recent funding announcement: {{Funding News}}. Write a one-sentence reason to book a demo call for a data analytics platform. Keep it respectful, not salesy.” <code>`</code></p>
<p>This yields unique, relevant lines for every row.</p>
<h3>Use Case 3: Lead Negation (Negative Filtering)</h3>
<p>Filter out noise. Use Nqzai to spot false positives.</p>
<p>Prompt:</p>
<p><code>`</code> “Does this company appear to be a reseller or agency rather than an end-user of AI tools? Answer only YES or NO based on their description: {{Company Description}}.” <code>`</code></p>
<p>Then filter out all rows where the response is “YES.”</p>
<h2>Realistic Trade-Offs and Considerations</h2>
<ul>
<li><strong>Cost per row:</strong> Each Nqzai API call costs a small amount of credits. Running 1,000 prospects through a three-step workflow (score, personalize, filter) uses roughly 3,000 API calls. Budget accordingly.</li>
<li><strong>Latency:</strong> Nqzai responses typically take 1-3 seconds per call. A table of 500 rows with three enrichment steps will take 20–30 minutes to process. Plan for this in batch runs, not real-time.</li>
<li><strong>Accuracy:</strong> Nqzai is powerful, but it is not perfect. Always validate a sample of 20 rows manually before deploying the output for outbound campaigns. AI models can hallucinate job titles or misinterpret industry context.</li>
<li><strong>Model choice:</strong> Nqzai offers multiple models. For simple scoring, use a faster/cheaper model. For complex personalization or analysis, use the higher-tier “pro” model.</li>
</ul>
<h2>Avoiding Common Mistakes</h2>
<ol>
<li><strong>Over-prompting:</strong> Do not ask Nqzai to do five things in one prompt. Split into separate steps. This improves accuracy and makes debugging easier.</li>
<li><strong>Ignoring rate limits:</strong> Clay has API rate limits depending on your plan. Add a 100ms delay between calls if you hit errors.</li>
<li><strong>No fallback:</strong> If Nqzai returns an error (e.g., for a row with missing data), set a default value (like “0” or “Error”) so your table doesn’t break.</li>
<li><strong>Forgetting to test:</strong> Run the workflow on 3–5 test rows first. Inspect the raw JSON response to ensure the parse step is extracting the correct field.</li>
</ol>
<h2>Key Takeaway</h2>
<p>Using Nqzai inside Clay turns a basic lead list into a decision-ready, AI-enriched pipeline. The real value is not just automation—it is the ability to <strong>apply nuanced judgment</strong> (fit, intent, personalization) to every single row without hiring an SDR to read 100 profiles.</p>
<p>To start: build a single scoring step on 20 test leads. Once validated, scale to your full list. The combination of Clay’s orchestration and Nqzai’s reasoning is one of the most practical AI lead gen workflows available today for B2B teams.</p>

# How to Use Nqzai With Clay for AI Lead Generation

Lead generation in 2024 demands more than static lists and manual enrichment. The combination of **Clay** (the go-to platform for multi-source data enrichment and outreach automation) with **Nqzai** (a specialized AI model for natural language querying and synthesis) creates a workflow that turns raw contact data into prioritized, action-ready leads—without requiring a data science team.

This article outlines a practical, step-by-step method for using Nqzai inside Clay to build a lead gen pipeline that is both scalable and precise.

---

## What You’re Building: The Core Workflow

The goal is simple: feed Clay a list of prospects (names, companies, titles) and have it return a scored, enriched set of leads—with ranked fit scores, personalized messaging angles, and cleaned data—all processed through Nqzai.

Clay handles the orchestration, the API calls, and the data plumbing. Nqzai provides the reasoning layer: analyzing a prospect’s LinkedIn bio, company Crunchbase description, or job posting to assess fit against your ICP (Ideal Customer Profile).

---

## Prerequisites

Before you begin, confirm you have:

- **A Clay account** (any tier that supports HTTP requests and API integrations)
- **An API key for Nqzai** (available from the Nqzai platform; note that it uses a standard OpenAI-compatible endpoint)
- A CSV or list of at least 10-20 test prospects (to avoid wasting credits on a broken workflow)

---

## Step-by-Step: Setting Up the Nqzai Integration in Clay

### 1. Create a New Clay Table

Start with a simple table containing the columns you want to enrich. At minimum:

- `First Name`
- `Last Name`
- `Company Name`
- `Company LinkedIn URL`

Upload your target list. Clay will auto-detect the columns.

### 2. Add the Nqzai API Call

Clay’s “Add Enrichment” menu includes a **“Send HTTP Request”** option. This is your gateway to Nqzai.

Configure the request as follows:

- **Method:** POST
- **URL:** `https://api.nqzai.com/v1/chat/completions` (or the endpoint provided in your Nqzai dashboard)
- **Headers:**
    - `Authorization: Bearer YOUR_NQZAI_API_KEY`
    - `Content-Type: application/json`
- **Body (JSON):** This is where the prompt lives.

Example body for a lead-scoring prompt:

```json
{
  "model": "nqzai-pro",
  "messages": [
    {
      "role": "system",
      "content": "You are a B2B lead scoring assistant. Evaluate the lead based on company industry, employee count, and job title. Score from 0-10 where 10 is a perfect fit for a SaaS company selling to HR tech."
    },
    {
      "role": "user",
      "content": "Company: {{Company Name}}. Industry: {{Industry}}. Title: {{Job Title}}. Employees: {{Employees}}. Score this lead."
    }
  ],
  "temperature": 0.2,
  "max_tokens": 150
}
```

**Important:** Use Clay’s “Merge Tags” (the curly-brace variables) to dynamically insert data from each row. This is what makes the workflow per-row.

### 3. Parse the Response

Nqzai returns a JSON object. You need to extract the score from it.

After the HTTP request step, **add a “Parse JSON” step** in Clay. Target the output field from the HTTP request (usually `result.data` or `output`). The actual content will be inside:

`choices[0].message.content`

Map that content to a new column, e.g., `Nqzai Score`.

If you want a numeric value, you may need a second parse step that extracts just the number (e.g., using a regular expression in Clay’s formula or a “Text Transform” step).

### 4. Enrich with Additional Context

A single score is useful, but Nqzai can do more. Create **additional HTTP steps** for:

- **Personalization:** “Write a 2-sentence outreach email opener referencing this lead’s recent [Company News].”
- **Intent signal extraction:** “Does this job posting suggest they are hiring for a role that uses our product? Answer YES or NO.”

Each step adds a new column of AI-processed intelligence.

---

## Three High-Impact Use Cases (With Nqzai)

### Use Case 1: ICP Scoring at Scale

Instead of manually reviewing 500 LinkedIn profiles, let Clay feed each profile’s `About Section` (pulled via a LinkedIn enrichment provider) to Nqzai. The prompt:

```
“Analyze this LinkedIn bio. Rate how likely this person is to purchase enterprise HR software (1-10). Consider job function, seniority, and company size. Output only the number.”
```

Result: a clean numeric column you can filter or sort.

### Use Case 2: Personalized Message Generation

Standard token personalization (e.g., “Hi {{First Name}}”) is table stakes. Nqzai enables **context-aware** personalization.

Prompt:

```
“Based on the prospect’s recent funding announcement: {{Funding News}}. Write a one-sentence reason to book a demo call for a data analytics platform. Keep it respectful, not salesy.”
```

This yields unique, relevant lines for every row.

### Use Case 3: Lead Negation (Negative Filtering)

Filter out noise. Use Nqzai to spot false positives.

Prompt:

```
“Does this company appear to be a reseller or agency rather than an end-user of AI tools? Answer only YES or NO based on their description: {{Company Description}}.”
```

Then filter out all rows where the response is “YES.”

---

## Realistic Trade-Offs and Considerations

- **Cost per row:** Each Nqzai API call costs a small amount of credits. Running 1,000 prospects through a three-step workflow (score, personalize, filter) uses roughly 3,000 API calls. Budget accordingly.
- **Latency:** Nqzai responses typically take 1-3 seconds per call. A table of 500 rows with three enrichment steps will take 20–30 minutes to process. Plan for this in batch runs, not real-time.
- **Accuracy:** Nqzai is powerful, but it is not perfect. Always validate a sample of 20 rows manually before deploying the output for outbound campaigns. AI models can hallucinate job titles or misinterpret industry context.
- **Model choice:** Nqzai offers multiple models. For simple scoring, use a faster/cheaper model. For complex personalization or analysis, use the higher-tier “pro” model.

---

## Avoiding Common Mistakes

1. **Over-prompting:** Do not ask Nqzai to do five things in one prompt. Split into separate steps. This improves accuracy and makes debugging easier.
2. **Ignoring rate limits:** Clay has API rate limits depending on your plan. Add a 100ms delay between calls if you hit errors.
3. **No fallback:** If Nqzai returns an error (e.g., for a row with missing data), set a default value (like “0” or “Error”) so your table doesn’t break.
4. **Forgetting to test:** Run the workflow on 3–5 test rows first. Inspect the raw JSON response to ensure the parse step is extracting the correct field.

---

## Key Takeaway

Using Nqzai inside Clay turns a basic lead list into a decision-ready, AI-enriched pipeline. The real value is not just automation—it is the ability to **apply nuanced judgment** (fit, intent, personalization) to every single row without hiring an SDR to read 100 profiles.

To start: build a single scoring step on 20 test leads. Once validated, scale to your full list. The combination of Clay’s orchestration and Nqzai’s reasoning is one of the most practical AI lead gen workflows available today for B2B teams.

#clay-ai

Stories tagged clay-ai

How to Use Nqzai With Clay for AI Lead Generation