We’re excited to bring AI into the Satuit solution. Satuit Agent, a new AI assistant built into SatuitCRM, is powered by our new MCP/AI layer, part of Release 18.4. This post covers the AI capability itself: why we built it, what it does, and how the rollout works. For the full rundown of everything else shipping in this release, see the v18.4 release notes.
Why We Built This
Investment management firms have told us the same thing in different ways over the past year: they want to use AI, but they can’t hand sensitive investor and client data to a generic AI tool that wasn’t built with compliance, permissions, or audit requirements in mind. Most AI assistants on the market are built for general use. They aren’t built for firms that need to know exactly who accessed what, when, and under what permission level.
Rather than build a closed AI feature that only works inside our own chat window, we built the MCP/AI layer on the Model Context Protocol (MCP), an open standard for secure, permissioned AI access. That decision means Satuit Agent’s underlying architecture can extend to other AI tools over time, under the same governance model, instead of locking firms into a single proprietary assistant.
The Benefits
The MCP/AI layer gives firms a way to bring AI into daily CRM work without giving up control over data access. Because it runs on an open standard, firms aren’t limited to a single vendor’s AI roadmap. Every interaction, whether through Satuit Agent or another connected tool, is governed by the same role-based permissions and audit logging that already apply to the rest of the CRM. Admins retain full visibility into what AI features are being used, by whom, and how often, so adoption doesn’t come at the cost of oversight.
How It Works
The MCP/AI layer sits underneath Satuit Agent and manages how AI tools connect to CRM data. Access is controlled through a dedicated AI Configuration page, where admins enable or disable AI access by record type, and a Rep Permissions diagnostic tool that shows exactly which functions each role can reach. All AI-driven interactions run through automatic HTML sanitization, and every change made through the assistant is logged the same way any other CRM change is. Token usage tracking and per-rep reporting give admins ongoing visibility into usage across the team. Full technical details are on the MCP/AI layer features page.
Getting Started
Test sites open on September 1, giving clients a sandbox environment to evaluate the MCP/AI layer and Satuit Agent before it touches production data. On September 28, the MCP/AI layer becomes generally available in production and ready for customers to purchase.
A few things to know before you plan your rollout: Satuit Agent is a paid, add-on feature and requires your firm to be on v18.4. It also requires the Satuit API to be enabled on your account, whether you choose Satuit-hosted AI or connect your own AI platform. Your Satuit representative can confirm what’s needed to get set up.
To get access to a test site starting September 1, contact Satuit. If you’d like to see the MCP/AI layer in action first, schedule a demo ahead of the September 28 launch.
