I Asked AI to Pull My February Sales Numbers. It Queried My ERP in Real Time.
No export. No dashboard. No BI tool. I typed a question in plain English, and an AI agent queried my FileMaker-based ERP system through a live API connection—returning driver-level commission breakdowns by week, with the data still warm.
This isn’t a proof of concept. This is production. And it’s running on a stack that most enterprise architects wouldn’t expect: @Claris FileMaker, @Anthropic’s Claude, and a protocol called MCP that quietly changes everything about how AI systems interact with business data.
Let me explain what we built, and why it points toward something much bigger.
The Problem Nobody Talks About
Most small and mid-market businesses run on systems that work. ERPs built in FileMaker, Access, legacy platforms—they manage complex operations and have years of institutional knowledge baked into their logic.
The industry narrative says these systems need to be replaced. Migrate to the cloud. Adopt a SaaS platform. Start over.
I disagree.
I run FMrug Software, a multi-tenant SaaS platform built on @Claris FileMaker that serves carpet and area rug cleaning businesses nationwide. It handles scheduling, dispatching, invoicing, payment processing, SMS communication, barcode-based inventory tracking, and mobile field operations. It’s not a toy—it’s the operational backbone for businesses that clean carpets and rugs for Fortune 500 companies and neighborhood customers alike.
The real question isn’t “how do I replace this?” It’s “how do I make what already works dramatically smarter?”
Enter MCP: The Protocol That Connects AI to Everything
Model Context Protocol (MCP), developed by @Anthropic, is an open standard that allows AI models like Claude to interact with external tools and data sources through a structured interface. Think of it as a universal adapter between AI and your business systems.
Here’s what that means in practice: instead of exporting data, building reports, or writing custom queries, you give AI a set of tools—and it figures out how to use them.
For FMrug Software, we built an MCP server that exposes our FileMaker database through a set of purpose-built tools:
- Query records with dynamic filtering, sorting, and field selection
- Inspect table schemas to understand data structure on the fly
- Count records with optional filters
- Retrieve individual records by primary key
The AI doesn’t receive a static data dump. It receives capabilities. It can explore the schema, understand the data model, and construct intelligent queries based on natural language questions.
When I asked “What are the current month’s sales?”—Claude examined the invoice table schema, identified the right date and total fields, constructed an OData v4 filter for February 2026, and returned 87 invoices with full breakdowns. When I followed up with “Calculate driver commissions at 30% by week”—it cross-referenced the driver table, matched driver IDs to names, segmented by week, and produced a formatted commission report.
No code. No SQL. No intermediate steps. Just a question and an answer grounded in live operational data.
Why FileMaker + OData Is the Quiet Powerhouse
Here’s something the enterprise world overlooks: @Claris FileMaker Server ships with a built-in OData v4 API. Out of the box. No middleware required.
That means any FileMaker solution—whether it’s been running for 2 years or 20—can expose its data through a standards-based REST API that any modern system can consume. Python, JavaScript, Go, AI agents—they all speak OData.
For businesses running on FileMaker, this is transformative. You don’t need to rebuild your system to make it AI-ready. You need to expose it properly and connect it to the right tools.
Our stack looks like this:
FileMaker Server → OData v4 API → MCP Server (Python) → Claude AI
The MCP server acts as an intelligent middleware layer. It handles authentication, translates natural language intent into structured OData queries, manages pagination, and formats responses for the AI model. FileMaker handles what it’s always handled—the business logic, data integrity, and operational workflow.
Each layer does what it does best. That’s not a compromise. That’s architecture.
What This Unlocks
The commission report was a simple example. Here’s where this gets interesting.
Agentic Customer Service. We’re building toward a system where an inbound customer text message triggers an AI agent that can pull up their complete service history, check scheduling availability, generate a quote, and book a job—all by querying the ERP in real time through MCP. The customer gets an instant, informed response. The CSR gets freed up for complex cases.
Conversational Business Intelligence. Instead of building dashboards that answer predetermined questions, you have an AI that can answer any question your data can support. “Which zone had the highest revenue last quarter?” “Show me customers who haven’t rebooked in 6 months.” “What’s our average ticket by driver?” These aren’t canned reports. They’re live queries shaped by context.
Multi-System Orchestration. MCP isn’t limited to one data source. The same AI session that queries FileMaker can also interact with marketing automation platforms, communication tools, and external APIs. We’re integrating GoHighLevel for marketing automation, Pipedream for workflow orchestration, and Twilio for SMS—all accessible to the same AI agent through MCP.
The Bigger Vision: NectoMax and the Connected ERP
Everything I’ve described so far is happening inside one vertical—carpet and rug cleaning. But the architecture isn’t vertical-specific. The pattern of FileMaker + OData + MCP + AI works for any service business running on a structured database.
That’s the thesis behind NectoMax.
NectoMax is my vision for what a connected ERP looks like for small and growing businesses navigating the critical growth phase from $500K to $5M in revenue. It’s the stage where operations get complex enough to demand enterprise-quality tools, but the budget doesn’t support enterprise-quality price tags.
The jump from $500K to $5M is where businesses either break through or break down. Scheduling gets harder. Customer communication gets messier. Financial visibility gets murkier. And the gap between what your larger competitors can afford in technology and what you can feels insurmountable.
NectoMax exists to close that gap.
The connected ERP isn’t a single monolithic platform. It’s an ecosystem of best-in-class tools—your operational database, your communication layer, your marketing automation, your AI—all connected through open protocols and APIs. MCP is the connective tissue. OData is the data layer. AI is the intelligence that makes it all accessible to a business owner who doesn’t have a CTO on staff.
What I’m building with FMrug Software is the proof of concept. A small business operator can ask a question in plain English and get the same caliber of operational intelligence that a company with a dedicated analytics team produces. Not because they hired a data engineer—but because their systems are connected in a way that makes intelligence emergent.
That’s the future I’m building toward. Enterprise-quality outcomes. Small business accessibility. No rip-and-replace required.
The Bigger Picture
I think we’re at an inflection point for small and mid-market businesses.
For decades, the technology gap between a company running FileMaker and a company running SAP felt enormous. AI and open protocols like MCP are compressing that gap rapidly. The businesses that will win aren’t the ones with the biggest tech budgets—they’re the ones that can connect their operational intelligence to AI systems fastest.
MCP makes that connection possible without ripping out what works. OData makes legacy systems accessible. And AI models like Claude from @Anthropic are sophisticated enough to reason about complex business data without hand-holding.
If you’re running a business on FileMaker—or any system with an API—you’re closer to AI-powered operations than you think. The path isn’t replacement. It’s integration.
Doug Hauenstein is the General Manager of Widmer’s Carpet & Rug Cleaning, founder of FMrug Software (fmrug.com)—a SaaS platform serving carpet and rug cleaning businesses nationwide—and the driving force behind NectoMax (nectomax.com), a vision for delivering enterprise-quality connected ERP tools to small and growing businesses. The MCP integration described in this article was developed collaboratively with @Anthropic’s Claude AI, which also serves as the AI engine powering the real-time query capabilities. FMrug Software is built on @Claris FileMaker.
Interested in the intersection of AI and operational systems for small business? Let’s connect.
