Published on August 18, 2025.
Over the past week, I’ve been reevaluating the role of SuiteAnalyzer in light of NetSuite’s newly announced AI Connector Service. My main concern has been whether data accessed through the Connector — especially when passed to AI models like Claude — is truly safe and compliant.
To get clarity, I asked Claude directly.
My question:
“What can you tell me about the policies for data storage, using data for training, and so on — especially when it comes to data that is made available to Claude via MCP tools?”
Claude’s response:
I can share what I know about Anthropic's general data practices, though I should note that for the most current and comprehensive information about MCP-specific policies, you'd want to check Anthropic's official documentation.
General Data Practices: Anthropic has stated that they don't use conversations with Claude to train their models without explicit permission. For regular Claude conversations, the data is typically retained for safety purposes but not used for training.
MCP Tool Context: MCP (Model Context Protocol) tools allow Claude to access external data sources and services during conversations. This raises some specific considerations:
• The data accessed through MCP tools (like your NetSuite data, Google Drive files, etc.) would generally fall under the same non-training policies as regular conversation data.
• However, the specific handling may depend on how the MCP integration is configured and what agreements are in place between you, the MCP tool provider, and Anthropic.
What I'd Recommend: Since MCP is a relatively new protocol and policies around it may be evolving, I'd suggest checking:
• Anthropic's Privacy Policy - https://www.anthropic.com/privacy
• The MCP documentation
• Your MCP tool provider's policies
The safest approach is to assume that any data you make available to Claude through MCP tools is subject to the same privacy considerations as data you'd directly share in a conversation, but you should verify the specific policies that apply to your use case.
Claude’s answer, combined with what I already know about Anthropic’s data practices, tells me a few important things:
1. No model training without explicit consent. That’s the same promise Anthropic makes about normal Claude conversations.
2. MCP data is handled under the same policy. The AI Connector doesn’t introduce a loophole where data suddenly becomes fair game for training.
3. Enterprises still need to validate both ends. Oracle/NetSuite handles the data extraction, Anthropic handles the analysis — compliance teams will want both policies in writing.
This means that using Claude with the NetSuite AI Connector is much safer and more compliant than I had initially feared. For most companies, it’s “safe enough” — and that changes the equation for SuiteAnalyzer.
SuiteAnalyzer’s value proposition was its compliance-first, offline analysis model. It ensured your data never left your controlled environment. But if Claude + the Connector already satisfy compliance and governance concerns, then SuiteAnalyzer’s differentiator shrinks dramatically.
That doesn’t mean SuiteAnalyzer has no future — there may still be niches where offline/air-gapped analysis is absolutely required. But for now, I’m putting active development of SuiteAnalyzer on pause.
While the Connector solves a lot, there’s still a glaring gap: it doesn’t provide access to NetSuite report data.
That’s significant. Many of the most valuable AI analyses I’ve generated — Income Statement summaries, AR/AP aging reviews, Inventory health checks — are powered by real NetSuite Reports, not raw record-level data.
And that’s where the opportunity lies.
While working on my first Custom Tool, a script that extends the NetSuite AI Connector, I discovered that the new SuiteScript 2.1 "Custom Tool" script type doesn't support the N/http, N/https, and N/search modules. That means that SuiteScript developers will not be able to make the results of Saved Searches available via the AI Connector.
And that's also significant, because a lot of businesses use saved searches extensively - for reporting, mostly. Without access to the data that those searches return, the AI Connector is potentially much less helpful to them.
I’m now shifting my focus to SuiteReports — a service that:
• Extracts NetSuite report data (Income Statement, Balance Sheet, Cash Flow, Trial Balance, and more).
• Refreshes it daily and stores it in your File Cabinet.
• Makes it instantly available to the NetSuite AI Connector (and therefore Claude and other MCP clients).
• Keeps all data inside your account and under your governance.
In addition to supporting NetSuite reports, SuiteReports can also be used to export the results of Saved Searches - and place the exports in a File Cabinet, alongside the report exports.
SuiteReports fills the Connector’s biggest gap and positions itself as an enabler of NetSuite’s AI ecosystem, not a replacement.
So here’s where I’ve landed:
• SuiteAnalyzer → Paused, except for niche offline use cases.
• SuiteReports → Full steam ahead, as the missing link between NetSuite reports, Saved Searches, and the AI Connector.
If Anthropic’s assurances about MCP data handling hold true, then the Connector will become the standard path for NetSuite AI analysis. SuiteReports will make sure the data those AI models need — your reports — is available, accurate, and governed.
Hello, I'm Tim Dietrich. I develop custom software for businesses that are running on NetSuite, including mobile apps, Web portals, Web APIs, and more.
I'm the developer of several popular NetSuite open source solutions, including the SuiteQL Query Tool, SuiteAPI, and more.
I founded SuiteStep, a NetSuite development studio, to provide custom software and AI solutions - and continue pushing the boundaries of what's possible on the NetSuite platform.
Copyright © 2025 Tim Dietrich.