At SuiteWorld 2025, Evan Goldberg stood on stage and said: "The biggest update of NetSuite since we founded the company." That's a bold claim for a product that's been in continuous development for 25+ years.
He was talking about NetSuite Next — the new AI-native interface that Oracle has been building in parallel with 2026.1's embedded AI features. Having spent time with both, I'll give you the honest take: some of it is transformative, some is incremental, and a few things are still vaporware waiting for a roadmap date.
This is the full breakdown — what's actually in 2026.1, what NetSuite Next actually is, what it means for SuiteScript developers, and which features deserve your attention now.
Let's start with the numbers because they tell a real story. In 2026.1, Oracle shipped 100+ AI agents embedded directly into NetSuite workflows — not add-ons, not optional modules, but features baked into the core product for all 43,000+ customers. Accounts payable automation adoption jumped from 7% to 29% in a single year — a 4× increase that represents a fundamental shift in how finance teams are using the product.
This isn't Oracle catching up to a trend. They've been building the infrastructure — the ML training pipelines, the anonymised benchmarking data from tens of thousands of customer accounts, the governance framework — for years. 2026.1 is where that investment starts arriving on feature lists.
What this means practically
Many 2026.1 AI features are enabled by default or require only a Feature Enable flag in Setup. The barrier to accessing them is low. The barrier to using them well — that's where this guide comes in.
Here's every significant AI-powered feature that shipped in 2026.1, with an honest assessment of each.
A real-time dashboard that tracks every task in your financial close, flags anomalies as they happen, and generates plain-English narrative summaries of what the numbers show. It ships with Exception Management (pattern-based anomaly detection) and Narrative Insights (generative AI summaries).
Verdict: The most immediately useful feature in 2026.1 for finance teams. The fraud detection alone — flagging vendor bank account changes before payment runs — is worth enabling. I've covered this in depth separately.
Deep-dive article available
See the full setup guide, real-world findings, and honest assessment of where it falls short in: NetSuite Intelligent Close Manager: Setup Guide, Real Impact & What Oracle Won't Tell You.
The bank reconciliation module now uses machine learning to automatically match imported bank transactions to NetSuite records. It learns from your corrections — the more you use it, the better it gets. For high-volume accounts (1,000+ transactions per period), this is a significant time saving. For small businesses with 50 transactions per month, it's nice but not transformative.
The matching confidence score is exposed in the UI, so your team can review low-confidence matches rather than treating all suggestions as automatic. That's the right design — AI-assisted, not fully autonomous.
Watch out for
The ML model trains on your specific account's patterns. If you have a mix of entities with very different transaction volumes, enable it on your highest-volume entity first to let the model train before rolling out to smaller ones.
If you're using NetSuite's Enterprise Performance Management module, 2026.1 adds AI-driven agents for budget planning and account reconciliation. The Planning Agent can generate budget drafts based on historical trends, seasonality patterns, and driver-based assumptions. The Reconciliation Agent automates the matching and sign-off workflow for balance sheet accounts.
These are EPM-specific, so they won't apply if you're not on that module. But if you are — and especially if you're managing a complex planning process across multiple cost centres — these agents represent a real workflow change, not just a UI polish.
This is the Text Enhance capability applied to financial reporting. After running a report — P&L, balance sheet, cash flow — a "Narrative Insights" button generates a plain-English summary of the significant movements, variances, and trends.
It works by reading the actual report data, not by hallucinating generic financial commentary. I've seen it produce surprisingly nuanced observations when the underlying data is rich. On a simple report with three line items, it's less useful.
Best use case: Management packs. If you produce monthly board reports or investor updates, having a first-draft variance commentary is genuinely time-saving — even if you rewrite 50% of it, the structural scaffolding is there.
A machine learning model that analyses historical sales data, margin patterns, and customer segments to suggest optimal pricing. It integrates with the existing Advanced Pricing module and generates recommended price points that can be reviewed and approved before applying.
This is genuinely novel territory for an ERP. NetSuite isn't the first to offer pricing ML, but it's the first to do it natively without a third-party integration. The quality of recommendations will depend heavily on how clean and consistent your historical pricing data is — garbage in, garbage out applies strongly here.
Data quality requirement
AI Advanced Pricing needs at least 12 months of clean transactional pricing history to produce reliable recommendations. If your pricing data has significant manual overrides or is inconsistently structured, clean it before enabling this.
A machine learning model that predicts when specific customers will actually pay their invoices, based on historical payment patterns. The predicted payment date appears on open invoices and can be used in cash flow planning.
The use case is cash forecasting — instead of assuming all Net 30 customers pay on day 30, your cash projection reflects the reality that your customer in segment A pays on day 22 and your customer in segment B pays on day 45. That accuracy compounds significantly when you're managing a portfolio of accounts.
An AI-synthesised customer profile that aggregates everything NetSuite knows about a customer — order history, service tickets, payment behaviour, communication cadence, renewal risk — into a single view. Think of it as a CRM summary that automatically updates as data flows through the system.
The insight layer is what makes this different from just a well-designed record view. Customer 360 surfaces patterns: "This customer's order frequency has dropped 40% in the last quarter" or "Three open support tickets are correlated with this customer's last three renewal cycles." It doesn't require a separate BI tool or custom reporting — it reads the data that's already in your account.
NetSuite Next is not a new product — it's a new interface layer that sits on top of the existing NetSuite platform. The same data, the same records, the same underlying logic — but presented through an AI-native UI built around conversational interaction, visual canvases, and autonomous agents that can take action on your behalf.
Oracle announced it at SuiteWorld in October 2025. The rollout is phased, with North American customers getting access mid-2026. The features below are what's been announced and demonstrated — some are available in preview, others are still on the roadmap.
A conversational AI assistant embedded directly in the NetSuite interface. You ask questions in natural language — "Show me all vendors with overdue invoices and the associated payment hold reasons" — and Ask Oracle queries the database, generates the results, and presents them without you building a saved search.
This is potentially the highest-impact feature for non-technical users. The saved search in NetSuite has a steep learning curve. Replacing it with natural language queries lowers the barrier to data access significantly. Whether the natural language parsing is reliable enough for complex multi-criteria queries is the open question — early demos have been impressive, real-world edge cases are still being discovered.
A drag-and-drop workspace where you can assemble data views, charts, AI summaries, and workflow actions into a personalised dashboard — without touching SuiteScript or custom records. Think of it as a cross between a low-code dashboard builder and an AI briefing document that refreshes automatically.
The AI layer means Canvas doesn't just display data — it can surface patterns, highlight anomalies, and generate contextual commentary inline. For power users who've historically relied on saved searches and custom dashboard portlets, this is a meaningful upgrade in the toolbox.
The most ambitious of the NetSuite Next features — an agent that can execute financial close tasks, not just track them. It monitors the close in progress, identifies tasks that can be automatically completed (certain reconciliations, accrual entries that match patterns), and executes them with configurable approval gates.
This is directionally where the 2026.1 Intelligent Close Manager is heading. The 2026.1 version tracks and surfaces. Autonomous Close acts. The distinction matters enormously from a controls perspective — and the design includes configurable human-in-the-loop checkpoints for that reason.
Not yet GA
Autonomous Close is in limited preview as of this writing. The GA timeline for broader availability is mid-to-late 2026. Don't plan a close process redesign around it until you have access to the preview.
The broader framework that underlies Autonomous Close — the ability to define multi-step workflows that AI agents can execute autonomously. An agentic workflow can span record creation, approval routing, data transformation, and third-party API calls, with the AI agent making decisions at each step based on configured rules and live context.
For developers, this is where the SuiteAgents Framework becomes relevant. Oracle is providing both a no-code interface for business users and a programmatic API for developers to build custom agents. The two layers interoperate — a developer-built custom agent can be surfaced through the same Agentic Workflows UI that a business analyst would use.
NetSuite Next also includes Document & Knowledge Integration — the ability to upload unstructured documents (contracts, supplier catalogues, compliance policies) and have AI agents reason over them alongside structured NetSuite data. This matters for procurement and legal workflows where the answer to a question spans a vendor contract and a purchase order simultaneously.
The natural question from anyone who writes SuiteScript: does this make my skills obsolete? The honest answer: no — but it changes what skills matter most.
NetSuite Next doesn't replace SuiteScript for complex business logic. It replaces simple automation — the kind of Scheduled Scripts that run nightly to push data between records, the User Event Scripts that fire on save to populate a field. Those use cases now have a lower-code alternative.
What NetSuite Next creates demand for: developers who understand how to architect AI-native workflows, who can build custom agents using the SuiteAgents Framework, and who understand the governance implications of autonomous decision-making in a financial system.
Oracle shipped several developer-facing AI capabilities alongside 2026.1 and NetSuite Next. Here's what's in the toolkit.
A native SuiteScript module that lets you call Oracle's hosted large language model from server-side scripts. You pass a prompt, optionally include context from the current record or search results, and receive a structured response. Governance cost: approximately 10 units per call, which adds up fast if you're calling it in a loop.
/**
* @NApiVersion 2.1
* @NScriptType UserEventScript
*/
define(['N/llm', 'N/record'], (llm, record) => {
const afterSubmit = (context) => {
// Only run on create/edit, skip delete
if (context.type === context.UserEventType.DELETE) return;
const rec = context.newRecord;
const memo = rec.getValue({ fieldId: 'memo' });
if (!memo) return;
// Classify the transaction memo into a predefined category
// Governance: ~10 units per llm.generate call
const result = llm.generate({
prompt: `Classify this transaction memo into one of: [payroll, vendor-payment, intercompany, other].
Memo: "${memo}"
Respond with only the category label.`,
model: llm.Model.ORACLE_GENAI,
});
rec.setValue({ fieldId: 'custbody_ai_category', value: result.text.trim() });
};
return { afterSubmit };
});Governance warning
N/llm calls count against your governance units. Do not call it inside forEach loops over large result sets — you will hit the 5,000-unit limit quickly. Batch your LLM calls or use a Map/Reduce script with careful unit budgeting.
Oracle's implementation of the Model Context Protocol — a standard interface that allows external AI agents (Claude, GPT-4, Gemini) to interact with NetSuite records and searches through a defined tool layer. Instead of building a custom REST integration for every AI use case, you configure the AI Connector Service once and any MCP-compatible AI client can access your NetSuite data with appropriate permissions.
This is significant for enterprise AI use cases where the business already has an AI platform investment (e.g., Microsoft Copilot, Claude for Enterprise) and wants to ground those models in live NetSuite data without a bespoke integration project.
A new script type — alongside the existing User Event, Map/Reduce, Scheduled, etc. — specifically for AI agent tools. A Custom Tool Script defines a capability that an AI agent can invoke: its inputs, outputs, and the SuiteScript logic that executes when the agent calls it. This is the programmatic layer beneath the Agentic Workflows UI.
/**
* @NApiVersion 2.1
* @NScriptType CustomToolScript
*
* This tool is callable by AI agents via the SuiteAgents Framework.
* It returns the open invoices for a given customer, formatted for
* consumption by an LLM context window.
*/
define(['N/search'], (search) => {
/**
* @param {Object} params
* @param {string} params.customerId - Internal ID of the customer
* @param {number} [params.limit=10] - Max results to return
*/
const execute = (params) => {
const { customerId, limit = 10 } = params;
const invoices = search.create({
type: search.Type.INVOICE,
filters: [
['entity', 'is', customerId],
'AND',
['status', 'is', 'open'],
],
columns: ['tranid', 'trandate', 'duedate', 'amountremaining'],
}).run().getRange({ start: 0, end: limit });
return invoices.map((r) => ({
invoiceNumber: r.getValue('tranid'),
date: r.getValue('trandate'),
dueDate: r.getValue('duedate'),
amountDue: r.getValue('amountremaining'),
}));
};
return { execute };
});The orchestration layer that coordinates custom agents, tool scripts, and data sources into coherent multi-step workflows. Think of it as a runtime that manages the agent loop: select a tool, call it, evaluate the result, decide on the next action, repeat until the task is complete.
The SuiteAgents Framework is available to SuiteApp developers and is the backbone of the SuiteApp.AI Marketplace — Oracle's curated directory of third-party AI-powered NetSuite extensions. If you're building a SuiteApp and want to expose AI capabilities to customers, this is the distribution path.
An AI coding assistant with full context of the NetSuite API — available as a VS Code extension and via the Cline AI interface. It can generate SuiteScript from a natural language description, explain governance implications inline, flag deprecated APIs, and suggest patterns based on the specific script type you're building.
Having tested it: it's good at boilerplate (search setup, record CRUD, error handling patterns) and less reliable on edge cases involving complex governance situations or nuanced record type behaviours. Use it to scaffold, then review carefully. It won't replace knowing the API — it will speed up developers who already do.
With this many features, the practical question is: where do I start? Here's the prioritised order I'd recommend for most organisations.
AI Advanced Pricing and Customer 360 are more complex enablements — they require data quality work before they deliver reliable results. Prioritise them in a second wave once the quick wins above are running smoothly.
2026.1 is the most substantial release Oracle has shipped since OneWorld became standard. The AI features aren't bolt-ons — they're built on years of ML investment using anonymised data from 43,000+ customer accounts. That training data advantage is something no third-party tool can easily replicate.
NetSuite Next, when it's fully available, is a genuine architectural shift. The conversational interface and agentic workflows will change how finance teams interact with the system day-to-day. The developer story is still evolving — the SuiteAgents Framework and Custom Tool Script type are promising but early.
My honest assessment: the 2026.1 embedded features are ready to deploy today and deliver real value. NetSuite Next is worth watching and planning for, but don't redesign your implementation around features that are still in preview. Enable what's GA, measure the impact, and revisit the roadmap items when they're stable.
One more thing
All the AI features in 2026.1 are included with standard NetSuite licensing — no separate AI add-on cost. Oracle is clearly betting on AI capabilities as the retention and expansion lever, not the per-feature monetisation model some competitors have chosen. For existing customers, that's a meaningful benefit.
NetSuite Intelligent Close Manager: Setup Guide, Real Impact & What Oracle Won't Tell You
Complete guide to NetSuite's 2026.1 Intelligent Close Manager. Full setup walkthrough, real-world findings from live deployments, and an honest assessment of where it falls short.
Scheduled Script vs Map/Reduce: A Decision Framework
When does Map/Reduce overhead hurt you? After running both in production on datasets from 100 to 500,000 records, here's the decision framework I use.