NetSuite has taken a major leap into the AI era by introducing SuiteScript Generative AI APIs, powered by the new N/llm module. For the first time, NetSuite developers can directly send prompts to large language models (LLMs), receive AI-generated responses, and integrate generative intelligence into SuiteScript automation.

This blog breaks down how the N/llm API works, what models are supported, usage modes, limitations, and real examples of how developers can use generative AI in their NetSuite scripts.

✔️ What Is the N/llm Module?

The N/llm module is NetSuite’s new SuiteScript interface that lets your script:

• Send prompts to an LLM
• Receive generated text responses
• Evaluate structured prompts (via Prompt Studio)
• Stream responses in real time
• Pass parameters & variables to AI models
• Work with Oracle-hosted generative AI safely

NetSuite integrates with OCI Generative AI Service, meaning your data stays within Oracle infrastructure—never sent to third-party providers for training.

This module is available only in SuiteScript 2.1 and only in supported regions.

🧠 How SuiteScript Generative AI APIs Work

The workflow is simple:

1. Your SuiteScript calls N/llm.generateText()
2. NetSuite forwards the request to OCI Generative AI
3. The LLM processes the prompt
4. OCI returns the AI-generated answer back to NetSuite
5. Your script continues with that result

All communication stays inside Oracle’s secure environment.

💡 If you don’t specify a model, NetSuite automatically uses Cohere Command R.

🚨 Important Considerations Before Using N/llm

• AI responses can be creative, not deterministic
• Always validate output before updating records or posting transactions
• Not available in all regions
• Governance limits apply (just like any SuiteScript call)

Generative AI should help automate tasks — not blindly control your data.

🔧 Supported LLM Models

You can set the model using the modelFamily parameter in:

• generateText()
• generateTextStreamed()

For Prompt Studio prompts, the model is chosen inside the prompt configuration.

Common model families include:

• cohere.command-r (default)
• cohere.command-r-plus
• meta.llama2
• custom configured OCI models

Each model supports different parameters (temperature, maxTokens, etc.).

🧪 Example: Using N/llm to Generate Text

/**
 * @NApiVersion 2.1
 */
define(['N/llm'], (llm) => {

    const example = () => {
        const response = llm.generateText({
            prompt: "Write a 2-line description for a new product: Wireless Smart Lamp.",
            modelFamily: llm.ModelFamily.COHERE_COMMAND_R,
            modelParameters: {
                temperature: 0.3,
                maxTokens: 100
            }
        });

        log.debug("AI Response", response.text);
    };

    return { example };
});

🧪 Example: Using Prompt Studio + N/llm.evaluatePrompt()

If you already created a custom prompt in Prompt Studio:

const result = llm.evaluatePrompt({
    promptId: 'custprompt_generate_sales_summary',
    variables: {
        customerName: "JADS Toys",
        startDate: "2024-01-01",
        endDate: "2024-12-31"
    }
});

log.debug("AI Result", result.text);

This is ideal for reusable business logic across multiple scripts.

🎛 Usage Modes (Free, On-Demand, Dedicated Cluster)

NetSuite offers three ways to use generative AI:

1. Free Mode (Limited Monthly Usage)

✔ No setup needed
✔ Ideal for testing or light usage

❌ Limited quota resets monthly
❌ Will throw an error when capacity is used up

Use this when experimenting or running small usage scripts.

2. On-Demand Mode (Paid, Unlimited via OCI)

Requires an OCI account + configuration.

Best for:

• Medium to high usage
• SuiteApps
• Use cases where you cannot afford quota limits

Charges usage per LLM inference request.

3. Dedicated AI Cluster (High Volume Production)

For enterprises running:

• Very high AI workloads
• Large SuiteApps
• High concurrency systems

This provides dedicated compute resources for maximum performance.

⚠️ SuiteScript Governance for N/llm

Each method has its own governance cost.

Examples:

• generateText() → costs 10 units
• generateTextStreamed() → costs 15 units
• evaluatePrompt() → varies

Always check the governance limits table when designing automations to prevent errors like SSS_USAGE_LIMIT_EXCEEDED.

💡 Real Use Cases for N/llm in SuiteScript

Here are practical ways NetSuite developers can use generative AI today:

1. Automate item description generation

Create clear, SEO-friendly item descriptions from basic attributes.

2. Summarize long text fields

Convert long notes or case messages into 1-2 sentence summaries.

3. Sales order insight extraction

Generate insights from transactions:

• Key delays
• Customer patterns
• Reorder recommendations

4. AI-powered Suitelets

Build chat-like assistance tools inside Suitelets.

5. Auto-generate emails, comments, or support responses

Instead of manually typing long updates.

6. AI-driven workflow decisions

Example: Interpret customer requests and route cases automatically.

🏁 Final Thoughts

The N/llm module is one of the most powerful additions to SuiteScript since the introduction of Map/Reduce. It brings true generative AI capability directly into NetSuite development — without relying on external servers or unsafe third-party APIs.

With this API, developers can now build:

• Smarter Suitelets
• Intelligent workflows
• Auto-generating descriptions
• AI-powered data processors
• Dynamic customer communications
• Custom enterprise-grade AI tools

AI in NetSuite is no longer a future concept — it’s ready to build with today.