PC Bennett’s Acumatica Insider Quarterly Report: Q2 2026

AI Studio + AI Assistant: What “Practical AI in ERP” Looks Like Now

Q2 2026 | Acumatica Insider Series

If you’ve been sitting in leadership meetings lately, you’ve probably noticed a pattern. Someone mentions AI. A few people nod knowingly. A few others quietly wonder what, exactly, they’re agreeing with. Then the meeting ends, and everyone goes back to their spreadsheets.

You’re not alone in that experience, and frankly, you’re asking the right questions.

This quarterly report is here to cut through the noise. We’ll translate Acumatica’s AI messaging into plain English, walk you through what AI Studio and the AI Assistant actually are, and show you where the practical value is today for your operations and finance teams. We’ll also address the governance and security questions you should absolutely be asking before you flip any switches.

First, Let’s Define “Embedded AI,” Because It’s Not What You Think

When Acumatica talks about “embedded AI,” they don’t mean a chatbot bolted onto the side of your ERP. They mean intelligence woven directly into the workflows you already use every day. That distinction matters enormously for both adoption and security.

Think about how you currently process an accounts payable invoice. Someone receives a PDF, manually keys data into Acumatica, and eventually a manager reviews and approves it. Embedded AI changes that sequence by meeting your team inside the existing screen rather than redirecting them to a separate tool or asking them to learn a new interface. The AI reads the PDF, populates the relevant fields, flags anomalies, and surfaces the information your approver actually needs to make a decision. Your AP clerk didn’t open a new tab. They just had a better Tuesday.

This design philosophy, working within your existing workflows rather than adding new ones, is what separates Acumatica’s AI strategy from the spray-and-pray approach you’ve probably seen from other software vendors. As Acumatica’s Chief Product Officer Jon Pollock put it at Summit 2026, everything announced is grounded in how users actually work. That’s not just good marketing copy. It’s a genuinely different engineering philosophy, and it shows up in how these tools are built.

What Is AI Studio, Really?

The simplest accurate description of AI Studio is this: it’s a low-code/no-code configuration framework that lets you, or your Acumatica partner, define exactly how AI assists specific business scenarios inside your ERP. It is not a feature you click on and use directly every day. It’s the engine under the hood that makes AI-powered actions possible, configured to your business’s specific context.

AI Studio works through a five-stage process. When a user triggers an action, either manually by clicking a button, automatically through a workflow, or eventually by an AI agent, the system gathers relevant data from the business document in context. That data populates a pre-defined prompt template, which gets sent securely to your chosen Large Language Model (LLM) provider. The response comes back, gets analyzed, and the relevant fields in your Acumatica document get updated accordingly. No data leaves your environment without your knowledge and your explicit configuration.

One of the most important things to understand about AI Studio right now is its security model. Acumatica operates on a “bring your own LLM” structure, meaning you choose which AI provider you connect to. Options currently include Anthropic, AWS, Azure, and OpenAI. You establish a direct relationship with that provider, set your own data-use terms, and hold your own API keys. Acumatica acts as the integration layer, transmitting data through encrypted API calls, but does not access, broker, or store any data exchanged between your system and your chosen LLM provider. For ops and finance leaders worried about where your vendor data is going, that’s your answer: it goes where you tell it to go, under terms you control.

What Is the AI Assistant? Think “Digital Co-Worker”

If AI Studio is the engine, the AI Assistant is the interface your team actually talks to. Acumatica described it at Summit 2026 as a “digital co-worker,” and that framing is genuinely useful.

Imagine your most experienced Acumatica user. They know where everything is. They can pull a customer’s order history from three months ago in about thirty seconds, tell you exactly which vendor invoices are pending approval, and find that one weird credit memo from Q3 without breaking a sweat. Now imagine that person is available to every member of your team, instantly, and never gets tired of answering questions.

That’s roughly what the AI Assistant does. Users ask questions in plain English, something like “What’s the current open AR balance for Coastal Supply Co.?” or “Show me all AP invoices over $10,000 pending approval,” and they get immediate answers pulled from live Acumatica data. No custom reports. No ticket to IT. No waiting until next week’s meeting. For finance teams who spend meaningful portions of their day hunting for information that’s technically in the system but hard to surface quickly, this is where the productivity math starts to get interesting.

The AI Assistant is also being positioned as the conversational front end for AI-powered workflows, including CRM activity summaries, sales opportunity reviews, and cross-sell suggestions. The AI Cross-Sell Assistant, for example, scans sales history for buying patterns, identifies items frequently purchased together, and ranks recommendations by relevance. For your sales ops team, that’s a meaningful upgrade from hoping your reps remember to check the customer history tab.

Separating Hype from Value: Where AI Actually Helps Today

Not every AI feature is equally mature, equally useful, or equally worth your attention right now. Here’s where we see the real, usable value for most Acumatica mid-market clients, and where we’d pump the brakes on expectations.

Accounts Payable Processing is probably the highest-ROI starting point for most organizations. Acumatica’s AP document recognition uses AI and machine learning to import PDF invoices directly from files or email attachments and transform them into AP documents automatically. The system can be configured to search for vendors by email address, match line items, and populate fields, dramatically reducing the manual keying your AP team does today. Combined with anomaly detection (more on that shortly), you get a one-two punch: faster processing and smarter exception flagging. This isn’t experimental or aspirational. It’s in production and in use.

Anomaly Detection is the feature that tends to quietly impress skeptics most. Rather than asking your finance team to manually scan long transaction lists looking for outliers, the system analyzes numeric fields across your Generic Inquiries, compares values against defined groups, and assigns severity levels to detected anomalies. Out-of-the-box, Acumatica ships pre-built inquiry templates covering sales order margin analysis, AP document costs, purchasing costs, and production variance, exactly the areas where errors and fraud tend to hide. Alaska Indoor Sports, highlighted at Summit 2025, used anomaly detection to completely rethink their AR collections process, separating genuinely late payments from those delayed by logistical realities specific to their Alaskan market. The productivity gain there wasn’t theoretical. It changed how a real team prioritized its daily work.

Customer and Vendor Data Hygiene is a quieter but meaningful use case. AI-assisted workflows can help surface duplicate vendor records, flag customers with inconsistent contact data, and identify accounts that haven’t had activity in meaningful time windows. For operations leaders who know their customer and vendor master data is messier than it should be (and that’s most of you, no judgment), this kind of ambient cleanup assistance is genuinely valuable without requiring a big-bang data project.

CRM Activity Summaries represent one of the most immediately compelling use cases for sales and customer success teams. Rather than asking reps to manually log detailed notes after every call, AI Studio can be configured to generate activity summaries, surface recent interaction history before a call, and flag accounts that are overdue for follow-up based on your defined engagement rules. The time savings are real, and the data quality tends to improve because summary generation is faster than manual logging.

Where we’d counsel patience: complex multi-module queries through the AI Assistant are still in development. AI Studio automation currently operates within the context of a single document and doesn’t yet access related entities in the same pass. User-defined fields, a gap noted in the 2025 R2 release, are slated to be addressed in 2026 R1. These aren’t reasons to avoid AI. They’re reasons to start with the right use cases rather than trying to solve everything at once.

How to Pilot Without Overcommitting

The single biggest mistake we see organizations make with new ERP features, AI or otherwise, is trying to roll out everything at once. The second biggest mistake is waiting until everything is perfect before starting anything. Here’s the approach that actually works.

Start with one high-volume, well-understood process. AP processing is the most common starting point for a reason: the volume is high, the current state is measurable (invoices processed per day, error rate, time to approval), and the improvement shows up quickly in metrics your CFO already tracks. You know exactly what good looks like before you start, which means you can evaluate whether the AI is actually helping without a lot of interpretive work.

Enable anomaly detection on one or two Generic Inquiries before broadening. Rather than turning on anomaly detection across every data set simultaneously, pick the area where your team currently spends the most manual review time. AP costs or sales order margins are good candidates. Run it in observation mode for a few weeks before acting on the alerts. This builds institutional comfort with how the system thinks and where it tends to be right versus where it needs calibration.

Treat AI Studio as a partner configuration task, not a DIY project. The tools are genuinely low-code, and Acumatica has done significant work to make prompt configuration accessible. That said, the difference between an AI workflow that saves your team an hour a day and one that creates new exceptions to manage is almost always in the prompt engineering and the testing. Work with your Acumatica partner, and PC Bennett specifically, to design, test, and validate automations before pushing them to your production environment.

What to Watch in 2026 R1

The upcoming 2026 R1 release, following Acumatica’s standard semi-annual cadence, represents a significant milestone for the AI features currently in experimental or preview status. Several important transitions are on the way.

AI Studio moves from experimental feature to full general release with 2026 R1. This matters for a few reasons. Support-level commitments are shifting from experimental to production. Additionally, previously noted limitations, including support for user-defined fields, are slated to be addressed. If you’ve been holding off on enabling AI Studio because you didn’t want to build on experimental functionality, 2026 R1 is your trigger.

Anomaly Detection is also transitioning in 2026 R1, with a rename and consolidation that unifies GL Anomaly Detection functionality into a single, simplified module across the platform. What previously required two separate configurations may consolidate into one, and the naming convention will be more intuitive for users who aren’t deep in the Acumatica product architecture.

The AI Assistant is expected to reach fuller availability, with expanded AI agent capabilities for customer service workflows and additional self-service portal options. The collaborative portals announced at Summit 2026, featuring universal login, in-product messaging, and AI-powered self-service, are also expected to land in 2026 R1. That changes the customer and vendor interaction model meaningfully for organizations that currently manage those relationships through email.

Keep your eye on the AI-powered reporting and insights enhancements as well. Improved out-of-the-box reporting paired with proactive anomaly detection in reporting outputs means your monthly financial review shouldn’t require someone to manually hunt for the unusual number anymore. The system will flag it and tell you where to look. For finance leaders who spend the first three days of every close cycle looking for the thing that doesn’t add up, this is the feature that quietly changes how month-end feels.

Governance, Security, and Rollout: The Questions You Should Be Asking

Before enabling any AI feature, responsible operations and finance leaders should be asking a specific set of questions. Vendors don’t always volunteer the answers, and the answers matter.

Where does my data go, and who controls it? Acumatica’s AI Studio architecture puts this control in your hands. You choose the LLM provider. You hold the API keys. You set the data-use terms with that provider. Acumatica transmits data securely but does not store, broker, or access it. That said, it is your responsibility to configure your chosen LLM provider’s data retention and privacy controls properly. This isn’t a “set it and forget it” situation. It requires intentional governance decisions on your part.

What data is being sent to the LLM, and what shouldn’t be? This is the governance question most organizations underestimate. AI Studio operates on the context of the current document, which may include vendor names, amounts, contract terms, or customer details. Before enabling automations, you’ll want to define what categories of data are appropriate to send externally and what should remain within your four walls. Your Acumatica partner should help you map this before configuration begins.

Who has permission to trigger AI actions, and who can modify AI configurations? AI Studio includes administrator-level controls for managing prompts and LLM connections, but permission design requires deliberate thought. The same principles that govern other sensitive ERP functions, least privilege access and segregation of duties, apply here.

How will you know if the AI is wrong? This is perhaps the most important operational question. Acumatica’s design philosophy keeps humans in the loop: current AI Studio automations are user-initiated, not fully autonomous, and responses can be reviewed before they update records. Even so, you should establish a clear process for reviewing AI-generated outputs, especially in the early weeks of any pilot, and define what an acceptable error rate looks like for your team before you start rather than after you encounter a problem.

How PC Bennett Can Help

PC Bennett has been working with Acumatica clients through multiple release cycles of AI feature development, and we bring a specific perspective that we think is genuinely useful right now. We’ve seen what works, what doesn’t, and where the hype-to-value gap is widest.

For clients who are AI-curious but not yet committed, we will work with you to assess best use cases and optimize security. It’s a practical conversation about your current workflows, your data quality, your governance posture, and where the highest-value starting points are for your specific business. This isn’t a sales exercise; it’s a diagnostic. You’ll leave with a prioritized list of use cases, an honest assessment of what’s ready versus what needs preparation, and a rollout roadmap that doesn’t overcommit your team.

For clients who are ready to move, we provide end-to-end configuration support for AI Studio workflows: prompt design, testing, validation, user training, and post-launch monitoring. We also help clients establish the LLM provider agreements, API configurations, and data governance frameworks that responsible AI deployment requires. The technical configuration and the governance design need to happen together, not sequentially.

As 2026 R1 approaches, now is exactly the right time to have this conversation. AI Studio will move to general availability, pricing will formalize, and the organizations that have already piloted and validated their use cases will be positioned to expand. Those that waited will be starting where you could be finishing.

Reach out to the PC Bennett team to schedule your AI Readiness Assessment. No hype, no pressure. Just a practical look at what Acumatica’s AI tools can actually do for your business, and what it takes to do it right.

---