Integrating AL Language in ERP Solutions

NetSuite just made it possible to have a conversation with your ERP NetSuite has adopted the Model Context Protocol (MCP), an open standard that enables secure interactions between AI models and data systems. The AI Connector Service is a protocol-driven integration that lets supported AI clients, including Claude and ChatGPT, directly access and interact with NetSuite data and functionality. The MCP Standard Tools SuiteApp provides tools that let you interact with your NetSuite data, including working with records, reports, saved searches, and SuiteQL queries, using natural language input. The architecture matters. The SuiteApp works with your existing NetSuite roles and permissions. It uses the same access controls as the NetSuite UI, so you can only see data and take actions allowed by your assigned roles. NetSuite explicitly supports "bring your own AI" with an extensible, protocol-based design, so you aren't locked into a single AI provider. This gives businesses long-term control over their AI strategy as models, platforms, and capabilities evolve. The MCP Standard Tools include customer management (create, update, search, and retrieve customer records, balances, and transactions), sales orders (retrieve orders, search with filters, get line item data), inventory (view item details and check inventory levels by location), and reporting (run SuiteQL queries, generate sales reports, view financial performance summaries). You describe the data you need in natural language, and the AI client automatically constructs and runs the appropriate query to retrieve it. The companies that build this into their operations first are collapsing the gap between business questions and ERP data.

In my previous articles in the "Bridging AI and ERPs" series, I explored how to connect Claude AI with SAP systems and how to meet organizations where they are without waiting for expensive ERP upgrades. Today, I'm sharing something practical: a working implementation that lets business users ask ERP questions in plain English. For finance teams, procurement, sales ops, and anyone stuck waiting for reports this changes the game: this solution can be deployed in hours, not years. No SQL, no BI team tickets, no waiting days to deliver values from your ERP data. I built and open-sourced a translator (MCP server) that connects Claude AI to Google BigQuery, where SAP data lives. The secret sauce is to leverage Google Cloud Cortex's pre-built semantic model, so you don't need to build the data logic from scratch. Business users ask questions in natural language, and Claude handles the SQL, queries BigQuery, and returns insights. Read the full article below. #AI #BigQuery #MCP #EnterpriseAI #SAP #DigitalTransformation #DataDemocratization #CloudComputing #BusinessIntelligence #FinanceAutomation #ERP #DataWarehouse #GenAI #Innovation #CFO #CIO

Stop Treating AI as a Side Project. Start Embedding It Into What You Already Build. Most teams think AI means “build a chatbot.” That’s the smallest use case. The real leverage comes when you embed AI directly inside your existing systems — ERP, CRM, gym management platforms, project management tools, SaaS dashboards, internal admin portals. AI shouldn’t replace your system. It should amplify it. Here’s where you can integrate AI in almost any existing software: 1. Intelligent Search & Context Awareness Instead of keyword search, integrate LLM-powered semantic search: “Show members who might churn next month” “Summarize this project status in 3 bullet points” “Find all invoices with unusual patterns” Use: OpenAI / Anthropic APIs Vector databases (Pinecone, Weaviate, PostgreSQL + pgvector) LangChain / Semantic Kernel for orchestration 2. Smart Reporting & Auto-Insights Your dashboards shouldn’t just show data. They should explain it. AI can: Detect anomalies Generate executive summaries Suggest actions Predict trends Imagine your admin panel saying: “Revenue dropped 12% due to lower renewals in tier-2 members.” That’s value. 3. AI Inside Forms & Workflows Instead of static input forms: Auto-fill fields from context Validate human-written descriptions Rewrite or optimize content Generate emails, notifications, reports automatically Every repetitive admin action is an AI opportunity. 4. Internal Knowledge Assistants Embed a private LLM trained on: Your documentation Your policies Your product manuals Your support tickets Now your team doesn’t search Slack history. They ask the system. 5. Process Automation with Intelligence Traditional automation follows rules. AI-based automation handles ambiguity. Combine: Message brokers (RabbitMQ, Kafka) Background workers (Celery, Hangfire) LLM APIs for decision making Now your system can decide, not just execute. Tools That Make It Practical Depending on your stack: Python ecosystem LangChain LlamaIndex FastAPI Celery pgvector .NET ecosystem Semantic Kernel Azure OpenAI Background Services / Hangfire MediatR pipelines for AI augmentation Frontend Streaming responses (WebSockets / SSE) Real-time AI feedback inside forms How to Start (Without Overengineering) Identify repetitive cognitive tasks (summaries, classification, tagging). Add AI as a feature, not a product. Start with one workflow. Measure time saved. Iterate. Don’t rebuild your architecture. Inject intelligence where it creates leverage. The companies that win won’t be the ones “using AI.” They’ll be the ones whose systems quietly think. If you’re building SaaS, internal tools, or enterprise systems, this is the moment to innovate inside your current product, not abandon it. AI isn’t the future. It’s an upgrade layer. Embed it.

Bringing AI and ERP systems together is not only possible but necessary for companies that want to lead with data and clarity. I have helped teams navigate this transition, and I know what works when you focus on alignment, purpose, and execution. If you are ready to make this shift successfully, here are five steps that will move you forward: 1. Create alignment across departments - Bring IT, operations, and leadership together in regular conversations. When teams understand the shared goal, they work faster and smarter. 2. Start with one clear use case - Choose a real business problem and apply AI and ERP tools to solve it. One strong result will build momentum and show the value quickly. 3. Prepare your data for decisions - AI is only as good as the data it receives. Take the time to clean and organize your ERP data so your insights are accurate and actionable. 4. Move from insight to execution - Do not let valuable insights sit unused. Create a process that ensures AI output leads to real decisions with clear ownership and follow-through. 5. Lead with clarity and purpose - Your team needs more than new tools. They need to understand why this matters. Share the vision and guide them with confidence and direction. This is how real change takes hold. To dig deeper into this approach, read more here: https://hubs.ly/Q03rzq_60

At Microsoft Build 2025, Microsoft introduced one of the most transformative steps for ERP and AI integration: the Dynamics 365 ERP Model Context Protocol (MCP) server. The MCP server is more than just a connector — it’s an open standard that defines how AI agents interact with enterprise data and business logic. Instead of relying on custom APIs or point-to-point integrations, MCP provides a unified, governed framework that standardizes access to ERP data and actions. This means agents, apps, and analytics can now securely access business data, perform transactions, and automate tasks directly within Dynamics 365 — with consistent context, permissions, and auditability. Why MCP matters The MCP framework: Enables agents to access data and business logic across multiple ERP apps Simplifies agent development and reuse across environments Provides consistent data governance and permissions Unlocks millions of ERP functions dynamically through form and API tools In practical terms, this brings us closer to agentic ERP : where AI doesn’t just analyze data but acts on it, securely and intelligently. Getting started: Configuring MCP in Dynamics 365 Finance and Operations (Preview) With the dynamic MCP server now available in public preview, you can start connecting your Dynamics 365 ERP environment to Microsoft Copilot Studio and other AI platforms. Here’s how to configure it step by step 👇 1️⃣ Check your environment Make sure your environment meets the prerequisites: Dynamics 365 Finance and Operations version 10.0.2428.15 or later Tier 2 or higher environment (CHE not supported) Unified Developer Environment recommended 2️⃣ Enable the feature In Feature Management, enable: (Preview) Dynamics 365 ERP Model Context Protocol Server 3️⃣ Allow MCP clients You must authorize the agent platforms that can connect to your MCP server. By default, the following are enabled: Microsoft Copilot Studio (Client ID: 7ab7862c-4c57-491e-8a45-d52a7e023983) GitHub Copilot (Client ID: aebc6443-996d-45c2-90f0-388ff96faa56) 4️⃣ Add the MCP server to your agent in Copilot Studio In  Microsoft Copilot Studio: Open your agent (or create a new one) Go to the Tools tab → Add a tool Filter by Model Context Protocol Select Dynamics 365 ERP MCP server Create a connection and Add to agent Your agent now has access to the ERP’s dynamic tools — including form navigation, action invocation, and data operations — based on its assigned security role.

🎙️ ES Con Microsoft Dynamics 365 Business Central ya no hablamos solo de automatizar procesos. En la nueva version 27.4 y en public preview ahora podemos crear agentes nativos en AL que navegan páginas, leen y escriben datos, toman decisiones y colaboran con los usuarios dentro del ERP. Nada de Power Platform. Nada externo. Todo dentro de Business Central. Hemos implementado nuestro primer agente para cualificar leads automáticamente y el impacto es claro: menos tareas repetitivas, más foco comercial y decisiones más rápidas. Esto no es una feature más. Es un cambio de paradigma para los desarrolladores AL. Nuestro Business Centra empieza a tener compañeros digitales. En este post no demasiado técnico os cuento con un caso de uso de que hablamos cuando nos referimos a custom agents (coding agents) 🎙️ EN With Microsoft Dynamics 365 Business Central, we are no longer just talking about automating processes. In the new version 27.4 and in public preview, we can now create native agents in AL that navigate pages, read and write data, make decisions, and collaborate with users within the ERP. No Power Platform. Nothing external. Everything within Business Central.  We have implemented our first agent to automatically qualify leads, and the impact is clear: fewer repetitive tasks, more commercial focus, and faster decisions.  This is not just another feature. It is a paradigm shift for AL developers. Our Business Central is starting to have digital colleagues.  In this not-too-technical post, I'll tell you about a use case that we talk about when we refer to custom agents (coding agents). #Dynamics365 #BusinessCentral #AI #Copilot #Agents #ALDC #PiensaEnGrande