SuiteWorld 2026 ran October 25–28 in Las Vegas, and NetSuite used it to make one argument: AI should do more of the work of running a business, inside the controls a company already has. Evan Goldberg called it “the end of ERP as we know it.” NetSuite now has more than 44,000 customers, and more than 15,000 of them already use AI with NetSuite in some form.
This recap is for NetSuite admins, finance leaders and business systems teams who want the substance without watching every session. We focus on what is available today, what is still coming, and the practical questions we keep hearing from NetSuite customers: how Ask Oracle really works, how fast the AI Units burn down, and whether you can run NetSuite from the AI tools your team already uses.
If you run accounts payable, our companion post covers what SuiteWorld 2026 means for AP teams.
Key takeaways
- AI reaches NetSuite two ways. From the outside, Claude, ChatGPT and other assistants connect through the AI Connector Service. From the inside, Ask Oracle and agents run in NetSuite. Most companies will use both.
- Ask Oracle now acts, not just answers. It creates draft records, builds saved searches and workflows, and shows the sources behind every answer. Nothing is saved until a person approves it.
- Agents come with governance. SuiteAgents run under a NetSuite role, can ask for approval, log their changes in System Notes and can be switched off with one click.
- The close is the flagship use case. Reconciliation agents, exception management and an intelligent close workspace aim to move finance from a month-end scramble to “always be closing.”
- The AI is included, but it has a meter. Native AI features draw on a pool of AI Units. Using NetSuite from Claude or ChatGPT through the AI Connector does not.
1. Two ways in: NetSuite from your AI, and AI inside NetSuite
The biggest shift at SuiteWorld 2026 is that NetSuite no longer expects every user to work in the NetSuite interface. NetSuite described two paths, “outside in” and “inside out,” and said plainly that most customers will use both.
Outside in: the AI Connector Service. The connector uses the Model Context Protocol (MCP) to let AI assistants work with NetSuite data and actions under your existing roles and permissions. NetSuite says more than 13,000 companies have used it, up from fewer than 1,000 a year ago. It is available in the Claude, ChatGPT and Grok marketplaces, with Copilot and Gemini named as coming soon. Customer examples in the keynotes were concrete. Every Man Jack reported processing three times as many transactions with the same team. Therabody showed a month-end close check that runs in Claude, finds the transactions blocking the close and routes them to the right team, with people approving each step.
SuiteSkills make those assistants smarter about NetSuite. They are more than 45 prebuilt skills across finance, operations, sales, projects and more, which teach any AI how NetSuite and common business processes work. In the keynote, the same receivables question asked in ChatGPT, Claude and Ask Oracle returned the same answer down to the dollar, because the expertise comes from NetSuite rather than from the model. SuiteSkills are free to download from GitHub and AI marketplaces.
Inside out: Ask Oracle and NetSuite Next. Ask Oracle is the AI assistant built into NetSuite. NetSuite says a few hundred businesses had tried it in July and more than 2,500 by September. It comes with NetSuite Next, the AI-first version of NetSuite, but you can also turn on Ask Oracle in the NetSuite version you use today.
For teams that asked whether NetSuite has a “headless” story, this is it. You can query, report on and update NetSuite from the assistant your people already live in, and still approve sensitive actions inside NetSuite. As NetSuite put it, the data, permissions and source of truth stay the same either way.
Connecting each employee’s own Claude or ChatGPT account to the ERP raises a fair governance question, though: who controls which models see finance data, what gets logged, and how spend is tracked? A private AI platform answers that. In our Ally Platform, OneChat gives employees Claude, GPT, Gemini and other approved models in one governed workspace, and it connects to any MCP server, including NetSuite’s AI Connector Service. Every NetSuite tool call runs under the same data masking, approvals and audit trail, so teams get headless access to NetSuite without each user wiring an ERP into a personal AI account.
2. Ask Oracle moves from answers to actions, with guardrails
The product keynote walked through Ask Oracle from simple questions to real work. What matters for most teams is how it stays controlled:
- It sees only what you can see. Ask Oracle works with the permissions of the person using it. One customer’s NetSuite admin tested this by switching between roles and confirming confidential data stayed hidden.
- It shows its work. Each answer lists the report, records and assumptions it used, and you can open the exact report or download the underlying data.
- It creates drafts, not final records. Drop in a quote or a customer PO and Ask Oracle creates a draft purchase order or sales order for review. Your customizations, approval workflows and SuiteScripts still apply when you submit it.
- It changes configuration only with approval. Ask Oracle can build saved searches and SuiteFlow workflows from a plain-language request, but an admin reviews and approves each one, and you control who may do this.
- Skills capture your way of working. Users and admins can save a repeatable task as an Ask Oracle skill, so the next person gets the same result without re-explaining it.
AI portlets bring the same capability to dashboards: you describe what you want to monitor, and Ask Oracle keeps it current on your home page.
3. Agents come with governance built in
NetSuite calls its agents SuiteAgents. Some come ready to use, such as industry agents for software, product and construction businesses. Others you build yourself. The keynote showed an agent created from a conversation in Ask Oracle, then refined in SuiteAgent Studio, where you set its goal, instructions, schedule and tools.
The controls are the part finance and IT teams should look at first:
- Every agent runs as a NetSuite role. Its access is exactly what that role allows, so your existing security model applies.
- People stay in the loop. A new layer NetSuite calls the agent trust protocol brings a person in when an action needs review.
- Changes are traceable. System Notes record which person or which agent made each change.
- Admins can monitor and stop agents. The SuiteAgent Management Workspace shows what each agent is doing and turns it off with one click.
One caution on timing: NetSuite’s press materials list SuiteAgent Studio, the SuiteAgent Management Workspace and the SuiteFlow Agent for a future release, while the product keynote demonstrated them as available. Analysts have noted the same gap. Check what your account shows before you plan around them.
4. The close is the flagship use case
If NetSuite has one showcase for AI, it is the financial close. The idea, in NetSuite’s words, is “always be closing”: catch and fix issues during the month instead of at month-end. The pieces shown in the keynotes:
- An intelligent close workspace that imports your existing close plan from a spreadsheet and tracks every task, whether a person or an agent owns it.
- Bank reconciliation agents that match transactions as they arrive and surface exceptions the same day.
- Balance sheet reconciliation for any account inside NetSuite, with an owner, sign-off and an audit trail auditors can see.
- Exception management that flags likely errors in real time and explains why.
- Flux analysis ready for review on day one of the new period.
Customers on stage reported real results. Johnny Pops said it cut its close nearly in half this year. Nature’s Emporium went from a 20-day close to three days. Both stressed the same lesson: the technology helped, but the process changes their teams made mattered just as much.
5. The AI is included, but it has a meter
Evan Goldberg closed the opening keynote by saying Ask Oracle, the AI Connector Service, SuiteSkills, AI portlets, agents and NetSuite Next itself are all included. That is true, with one detail worth understanding before you roll AI out widely: NetSuite’s built-in AI features consume AI Units.
According to NetSuite’s AI Units FAQ:
- Each general access user adds 1,000 AI Units per month of your term to one account-wide pool, provisioned upfront for the whole term. Ten users on a 36-month term get 360,000 units.
- Units expire at the end of the term and don’t carry over.
- When the pool runs out, you buy add-on packs of 10,000 or 25,000 units or pause AI features. NetSuite hasn’t published add-on pricing.
- Admins track usage under Setup > Company > View Billing Information.
- The AI Connector Service does not consume AI Units, and neither do Bill Capture, NetSuite EPM or NetSuite Analytics Warehouse.
NetSuite’s usage estimates give a sense of scale:
| Activity | Estimated AI Units |
|---|---|
| Simple Ask Oracle question | About 10 |
| Ask Oracle analysis question | 10–50 |
| Ask Oracle research question | 50–200 |
| SuiteScript N/llm request | 5–25 |
| Narrative Insights | 5–200 |
NetSuite hasn’t published estimates for agents, AI portlets or workspaces, which run repeatedly in the background. That is where consumption is hardest to predict, so pilot an agent and watch the billing page before you switch it on across the company.
Also announced
- Ask Oracle predictions, coming soon: forecasts of revenue, margins or inventory from your own history.
- AI workspaces, one place to run a whole process with people and agents, such as the close workspace above. The AI Workspace Builder, coming soon, will let you create your own from a plain-language description.
- Ask Oracle Connectors, coming soon: Ask Oracle reaching the web and other business applications, such as a forecast in SharePoint.
- NetSuite Data Intelligence, the new name for NetSuite Analytics Warehouse, with a Data Intelligence agent in Ask Oracle starting in November.
- SuiteExpress, an AI-assisted implementation toolset. One early customer went live in 60 days against an expected six months.
- QuickStart, a free NetSuite offering for early-stage startups.
- Developer tools: the SuiteCloud Developer Assistant, and SuiteCloud Agent Skills that teach AI coding assistants NetSuite best practices.
What NetSuite teams should do now
- Turn on Ask Oracle for a pilot group. You don’t need to move everyone to NetSuite Next. Start with reports people already trust, so they can check the answers.
- Connect one external assistant. If your team already uses Claude or ChatGPT, set up the AI Connector with a dedicated role and test read-only reporting first.
- Review roles before you add AI. Both Ask Oracle and the connector inherit role permissions. Over-broad roles become over-broad AI.
- Clean up the data AI depends on. Vendor and customer records, GL segmentation and item data drive the quality of every answer.
- Budget AI Units. Estimate usage for heavy users and any background agents, and check the billing page weekly at first.
- Write down your approval rules. Agents enforce the boundaries you give them, so decide what they may do alone.
Boston SoftDesign helps companies put NetSuite AI to work safely, from Ask Oracle rollouts to Claude and ChatGPT integrations over MCP and custom agents. Explore our NetSuite services, our AI consulting services, or talk to our team.
Frequently asked questions
When and where was SuiteWorld 2026? SuiteWorld 2026 took place October 25–28, 2026, in Las Vegas. Keynotes and many sessions were also streamed through SuiteWorld On Air.
Do I need NetSuite Next to use Ask Oracle? No. You can turn on Ask Oracle in your current NetSuite account, switch to NetSuite Next, or try both in a preview account. Admins can enable NetSuite Next role by role or user by user, with no reimplementation required.
Is AI in NetSuite free? NetSuite includes Ask Oracle, AI portlets, agents, SuiteSkills and the AI Connector Service with NetSuite Next. Built-in AI features draw on a pool of AI Units that comes with your user licenses; if you use them up, you buy add-on units or pause AI use.
How many AI Units do we get, and what happens when they run out? Each general access user contributes 1,000 AI Units per month of your term, provisioned upfront into one account-wide pool that expires at the end of the term. When the pool runs out, you can buy add-on packs of 10,000 or 25,000 units or pause AI features until the next term.
How many AI Units does an Ask Oracle question use? NetSuite estimates about 10 units for a simple question, 10–50 for an analysis question and 50–200 for a research question. Actual usage depends on the complexity of the request.
Can I use NetSuite from Claude, ChatGPT or another AI assistant? Yes. The NetSuite AI Connector Service uses the Model Context Protocol (MCP) to connect AI assistants to NetSuite. It is available in the Claude, ChatGPT and Grok marketplaces, and any MCP-compatible client can connect. Users can query data, build reports and create or update records within their role’s permissions. Companies that want central governance can connect it through a private AI platform such as Ally, whose OneChat works with any MCP server under one set of approvals and audit logs.
What do we need to set up the AI Connector Service? Enable REST web services, server-side SuiteScript and OAuth 2.0, install the MCP Standard Tools SuiteApp, and create a dedicated role with the MCP Server Connection, REST Web Services and OAuth 2.0 access token permissions. The Administrator role can’t be used to connect.
Does the AI Connector Service use AI Units? No. According to NetSuite, the AI Connector Service does not consume AI Units. You pay for your own AI assistant, such as a Claude or ChatGPT plan, instead.
Can Ask Oracle or the connector see data my role can’t? No. Ask Oracle uses the permissions of the person using it, and the AI Connector Service runs under the role used to connect. SuiteAgents run under a role you assign.
How do I see what AI changed in NetSuite? System Notes record changes made through Ask Oracle, agents and the AI Connector, including whether a person or an agent started them. For the connector, the integration record’s execution log also shows each tool call and any errors.
What are SuiteSkills? SuiteSkills are more than 45 prebuilt skills that teach AI assistants how NetSuite and common business processes work, across finance, operations, sales, projects and more. They work in Ask Oracle and in external assistants connected through the AI Connector Service, and are free to download from GitHub and AI marketplaces.
Will agents change records without approval? Only within the boundaries you set. SuiteAgents run under a NetSuite role, can require a person to review an action, log every change in System Notes and can be switched off with one click. Ask Oracle creates draft records for review rather than final records.
Where is NetSuite Next available? NetSuite Next is live in the US and Canada. The rollout is starting in the UK, Australia and New Zealand, with other EMEA markets from 2027.