Sage Intacct\u2019s connected finance push now raises a harder follow-up question: how much access should AI have to the finance system?<\/strong><\/p>\n\n\n\n ERP Today covered Sage\u2019s move to connect planning, expense management, receivables, cash flow and industry workflows inside Sage Intacct. The next layer is less about adding more finance functions and more about governing the data and controls AI will need to operate across them.<\/p>\n\n\n\n That is where Sage Intacct 2026 R2 becomes<\/a> more interesting than a standard release update. The release brings AI Gateway into the Intacct architecture, adds intelligent three-way matching in AP Automation, and continues Sage\u2019s push into construction and distribution workflows. Together, these moves show how mid-market finance AI is moving from embedded assistance towards controlled access to live ERP data.<\/p>\n\n\n\n For CFOs, that shift cuts both ways. AI can reduce manual reconciliation, surface exceptions more quickly and support more responsive finance operations. It can also create new exposure if data access, permissions, audit trails and workflow boundaries are not designed before automation expands.<\/p>\n\n\n\n Sage Intacct AI Gateway is the clearest signal in the release because it changes where finance AI is allowed to connect.<\/p>\n\n\n\n Sage describes AI Gateway as a secure, standardised bridge between Sage Intacct financial data and AI applications, giving customers and partners access through REST APIs and the Sage Intacct Model Context Protocol (MCP) server<\/a>. The MCP server is designed for AI interactions and read-only access, while REST APIs provide broader<\/a> access under the roles and permissions defined in Intacct.<\/p>\n\n\n\n That distinction matters for finance teams. AI assistants and external applications need context to be useful, but finance data is not a general knowledge base. It includes bank activity, vendor records, customer balances, invoices, approvals, dimensions, project costs, payroll and close data. Access to that information needs to respect the same permissions and control logic that govern users inside the ERP system.<\/p>\n\n\n\n AI Gateway gives Sage a way to support external AI workflows without treating Intacct data as an unmanaged export. The architecture allows customers to build AI processes around their own finance data while keeping access tied to defined roles, permissions and governance. For a mid-market customer base that may not have large internal AI engineering or security teams, that control layer becomes central to adoption.<\/p>\n\n\n\n The risk lies in implementation discipline. A gateway can standardise access, but it does not decide which AI workflows should exist, which users should approve them, or which actions should remain read-only. CFOs and CIOs will need governance models that define where AI can query, where it can recommend and where it can trigger action.<\/p>\n\n\n\nAI Gateway Moves the Control Point<\/strong><\/h2>\n\n\n\n