Your team is probably already using AI assistants, just not with your real numbers. Analysts copy figures out of the ERP, paste them into a chat window, and hope nothing sensitive went along. For most finance leaders, the obstacle to AI in finance was never doubt about the technology. It was a harder question: how do you give AI access to company data without losing control of it?
The CFO's Guide argues that the answer arrived as a protocol, not a smarter model. Written by Glenn Hopper, Head of AI and Managing Director at VAi Consulting, and published by Oracle NetSuite, it explains the Model Context Protocol (MCP) in plain business terms. It also shows what MCP changes for finance teams that have held back for good reasons.
Why the Model Context Protocol changes AI in finance for CFOs
The guide starts by taking finance caution seriously. It draws on recent research to explain why so many CFOs see real potential in AI yet keep it away from core accounting work. The worries are familiar: fragmented data, thin technical skills on the team, and no formal rules for how AI should be used. The author treats these as legitimate concerns, not resistance to change.
From there, it explains what MCP actually does. Before the protocol, connecting each AI tool to each business system meant separate custom integration work, which was expensive, fragile and hard to govern. MCP replaces that with a standard connection that respects the roles and permissions you already maintain. The core argument is simple. You don't have to just "trust AI." You can extend the trust model you've already built, so the AI sees only what a given user is allowed to see.
Finance AI use cases and AI governance you can explain to auditors
Abstract benefits don't get budget approved, so the guide grounds the idea in four everyday workflows across accounts receivable, accounts payable and FP&A. Each one contrasts today's manual routine with what changes when an assistant can query live ERP data, and each points to published research on the potential impact.
Governance gets equal attention. You'll see how existing SOX and SOC 2 obligations apply when AI supports financial reporting, what your audit trails should record, and where regulators currently stand. The guide also explains the design choices behind the NetSuite AI Connector Service, including the restrictions built in by default. It closes this part with a short comparison of protocol-based and embedded AI approaches among major ERP vendors, so you understand the tradeoffs before you commit.
A 90-day AI pilot framework plus a NetSuite MCP setup guide
Most AI pilots fail quietly: too many use cases, no executive sponsor, results nobody can translate into dollars. The guide lays out a three-phase, 90-day pilot framework designed to avoid that. It covers how to choose a first use case, who belongs on the pilot team (skeptics included), which metrics to track, how to set a budget ceiling, and how to make a clear go/no-go decision at the end.
The second half is written for the person who will do the work. It's a practical technical guide for NetSuite administrators implementing the MCP Standard Tools SuiteApp. It's specific enough that your admin can work from it directly, and it's clear enough that you'll understand the decisions being made on your behalf.
- How the client and server architecture works, explained in about five minutes
- Why a dedicated, least-privilege role is your primary control, and the configuration mistakes to avoid
- The available tool categories, their known limitations, and how to explain them to finance users
- Security guardrails, credential management, and migration guidance for early adopters
- Troubleshooting help and a go-live checklist
Who will get the most from it
CFOs and controllers building a case for AI will find the strategic framing and pilot plan most useful. FP&A, AR and AP leaders will recognize their own workflows in the use cases. NetSuite administrators get a setup reference they can follow. The full read takes about 13 minutes.
Download the full guide to get the use cases with their supporting research, the audit trail requirements, the complete 90-day pilot framework with target metrics and budget guidance, and the step-by-step NetSuite MCP setup and go-live checklist. It's everything you need to move AI from side experiments into governed finance work.