NetSuite Separates AI Access From AI Choice; That Is Worth Attention in Any ERP Evaluation
Every ERP vendor is now promising AI. The real question for a CIO is not whether the ERP has AI features, because they all do, but whether those features lock you into that vendor’s AI platform or let you use tools your organization already has. At SuiteConnect London in March 2026, NetSuite gave a clear answer to that question. This brief explains what they announced, what it means practically, and what to press on before you sign.
NetSuite is a cloud ERP platform owned by Oracle that handles finance, accounting, inventory, and operations for over 43,000 organizations. Like every major ERP vendor right now, NetSuite is adding AI. Instead of building a proprietary AI assistant and pushing customers toward it, NetSuite is letting organizations connect AI tools they already use directly to their NetSuite data. This reduces agent sprawl, an issue Info-Tech Research Group members are reporting more often. We have seen this as a good way for customers who may be in different stages of agent development themselves.
Microsoft Copilot, Claude, or something an IT team evaluated in the past year may be the agents that organizations are already running. Adding an ERP that comes with its own AI assistant, one that only works inside that ERP, creates training overhead and additional licensing cost and results in staff toggling between disconnected systems. The NetSuite approach avoids that by using a shared integration standard called the Model Context Protocol, or MCP. Think of MCP as a USB port for AI: one standard interface that any compatible AI tool can plug into. NetSuite built its AI layer on top of MCP, so the AI your organization already uses can connect to NetSuite data without custom integration work. The AI model is your choice. The access controls stay in NetSuite.
AI Connector Service Companion
Once the AI is connected to ERP data, organizations must address its lack of understanding of their specific finance structures. Generic AI does not distinguish between parent entities and subsidiaries or recognize unique cash categories used by different teams. Relying on it for finance questions often results in inaccurate, overly confident answers.
NetSuite Companion addresses this with three components.
- A library of over 100 prebuilt question templates, organized by role, covers common finance tasks for CFOs, controllers, AP analysts, AR analysts, and treasury analysts.
- Companion Skills are instruction sets that tell the AI how NetSuite’s data is organized, so outputs reflect actual data structures rather than a generic interpretation.
- MCP-ready roles enforce access automatically: An AR analyst using the AI sees only what that role is permitted to see in NetSuite, nothing more. The AI cannot become a shortcut to data an employee would not otherwise be allowed to access.
For a CIO, the governance point matters more than the templates. Role-scoped AI access means you can approve AI use in finance workflows without opening a new data exposure risk. Whether the templates are accurate enough for your organization’s specific setup is something to test before rollout, not after.
NetSuite MCP Apps
Getting useful answers from AI usually requires knowing how to write a good question. Many finance staff will not have the skills for this, and they should not have to. MCP Apps put familiar NetSuite menus and filters directly inside supported AI assistants. Staff select from dropdowns and pickers they already recognize from NetSuite, rather than typing free-form queries. This means fewer wrong answers, less support overhead, no retraining required.
MCP Apps are not yet released as of March 2026. NetSuite has announced they are coming through the SuiteApp Marketplace but has not confirmed the date.
AI Connector Service for NetSuite Analytics Warehouse
Until now, the AI Connector Service only worked with live transactional data: open invoices, current balances, active purchase orders. The Analytics Warehouse holds historical data, multiyear trends, and data from other systems your organization uses. This extension brings that data into the same AI interface under the same access controls.
The practical value is that finance staff can ask forecasting or trend questions and get answers from real historical data, without asking an analyst to first export and combine data sets manually. For organizations operating across multiple countries or legal entities, confirm with NetSuite how data residency requirements are handled in the Analytics Warehouse before enabling this.
Our Take
The AI lock-in question is one you should be asking every ERP vendor on your shortlist, not just NetSuite. Ask specifically: If we decide to change AI tools in two years, what happens to our ERP integration? Some vendors will tell you their AI is inseparable from the platform. NetSuite’s integration layer uses an open standard that supports any compatible AI, making it a practical architectural decision rather than a marketing claim. This has significant switching costs, especially as ERP systems typically last for many years.
The Companion is where near-term value will be tested. Finance teams are not skeptical of AI in the abstract; they are skeptical because they have seen it produce confident wrong answers. A controller who presents an AI-generated cash position to the CFO that turns out to be pulling from the wrong entity does not just have a data problem – that is the kind of mistake that ends careers. The role-scoping and prebuilt templates address that directly. Ask NetSuite to demo the Companion against a realistic data set from your industry before you accept that output quality meets your standard.
One thing this announcement does not fix: messy user roles. The AI sees what the role permits. If roles in your NetSuite environment are overpermissioned, outdated, or shared across users, the AI inherits that problem. Before enabling any of these features, a role audit is the right first step. It is not glamorous work, but it is the prerequisite. Build it into your implementation timeline.
Four questions to put to NetSuite before signing: Which AI tools are certified to work with MCP today, not on the roadmap? What is the confirmed release date for MCP Apps through the SuiteApp Marketplace? Can we run a pilot of the Companion against our own data during the evaluation? And for multi-entity or multi-country deployments: how does Analytics Warehouse AI access handle data residency requirements?