NetSuite and Acumatica made opposite strategic bets on AI in their 2026 releases. Each fits a different kind of mid-market buyer.
This is The ERP Brief. Weekly, for mid-market ERP buyers, operators, and the consultants who serve them. Patterns from real conversations. Honest analysis of what vendors ship and how to apply it. Sai Irrinki writing.
What’s new
NetSuite 2026.1 (Feb) and Acumatica 2026 R1 (Mar) both shipped major AI features. NetSuite emphasized openness: Narrative Insights for financial commentary, Intelligent Close Manager, CPQ AI Assistant, and MCP support so third-party AI tools can query NetSuite data directly. Acumatica emphasized first-party governance: an AI Assistant for plain-language queries, AI Studio for prompt governance, Data Masking, and AI Usage Tracking with token-level cost visibility. Two thoughtful product strategies pointed in different directions.
Why it matters for mid-market
NetSuite’s bet is interoperability. MCP support lets external AI tools touch ERP data directly. That is powerful for teams with AI engineering capacity or a strong analytics partner who can build the overlay. The strategy assumes your AI direction is best owned by your team, not by your ERP vendor.
Acumatica’s bet is built-in trust. Data Masking ensures confidential data never leaves the ERP boundary during AI use. AI Usage Tracking gives finance and IT line-item visibility into LLM cost per user and per feature. The strategy is built for teams whose AI roadmap is gated on governance before capability.
Both bets make sense. The right one depends on where your buying team sits. Teams with AI engineering or analytics partners gain the most from the open path. Teams whose CFO and CISO have held the line on AI in finance gain the most from the governance path.
A first-generation reality cuts across both vendors and the broader category: AI in ERP is at the productivity-assist stage, not the autonomous-action stage. First-generation drafting tools produce text that controllers edit substantially because models lack the business context that lives in Slack, email, and CRM notes. AI assistants work well on well-formed questions and need help on ambiguous business-language queries. Anyone in the market pitching “AI agents that run your finance ops” is overselling. Set expectations accordingly.
What to do this week
ERP Operations: If your team has AI engineering capacity or a strong analytics partner, scope the first AI workflow you would wire to your ERP via MCP. Pick something narrow with a measurable outcome, like AP exception classification or close commentary drafting. 2 to 3 hours to scope.
Finance teams: If your CFO or CISO has blocked AI in finance on data residency or unpredictable cost, schedule a 30-minute demo of governance features in your platform’s most recent release (Acumatica Data Masking and AI Usage Tracking, or NetSuite MCP if you have an AI engineering partner). 30 minutes plus the demo. The vendor conversation has moved and your prior blocker may now be removable.
All: If your back-office is wobbling two years post-implementation, the fix is rarely a re-implementation. A Solution Pack scopes the specific things slowing operations, ranked by ROI. Reply PACK for a sample. 1 minute to reply.
Steal This
A 5-item AI-in-ERP governance audit. 30 minutes. Pin this for your next CFO/CISO review.
List every AI feature currently active or proposed in your ERP. Include built-in (Narrative Insights, AI Assistant) and third-party (MCP overlays, voice agents, AP automation).
For each, document: data scope, data residency, authentication method, cost model, and named owner.
Score each on a Governance Risk scale: green (data masked, predictable cost, single owner), yellow (one factor unknown), red (data leaves boundary OR unpredictable cost OR no owner).
Take the audit to your CISO and CFO. Green and yellow items get approved. Red items get remediated or paused.
Re-run quarterly. New AI features ship every quarter and the audit is your governance pace layer.
One number
6 days. Time to build a $35M company’s full Business Requirements Document, Solution Design, and Proposal last week. Same scope, three years ago: 6 weeks.
POV
AI in ERP is at the productivity-assist stage. The buyers who win will set expectations there and skip the agent hype.
Reply
What is one AI feature in your ERP roadmap that has not lived up to the marketing? Hit reply. I synthesize patterns in next week’s issue, anonymized.
Sai Irrinki, Founder, TurnkeyERP
About TurnkeyERP: We deliver productized ERP work for mid-market companies via the AI Delivery Chain. Synchronized BRD, Solution Design, and Proposal generated by AI and reviewed by senior NetSuite, Acumatica, Oracle Fusion, and Oracle EBS consultants.
Forward to one ERP buyer or consultant who’d want this. Reply PACK for a sample.