The MBS FileMaker Plugin provides a comprehensive set of components to integrate Artificial Intelligence (AI) and Machine Learning (ML) into FileMaker solutions. This includes local LLM execution, platform-native ML frameworks, external API integration, and scripting bridges to other AI ecosystems.
Below is a technical overview of the available components and how developers can use them in real-world scenarios.
Foundation Models
Provides access to Apple's Foundation Models framework for running on-device large language models (LLMs).
-
Initialize and manage model sessions
-
Send prompts and receive streamed or complete responses
-
Configure parameters such as temperature, token limits, and system prompts
-
Supports multilingual input/output
Typical use cases include chat interfaces, text summarization, structured data extraction, and natural language command parsing—all without requiring internet access.
See FoundationModels functions and Use Foundation Models in FileMaker.
Llama.cpp
Integrates llama.cpp to run open-source LLMs locally within FileMaker.
-
Load GGUF/GGML models from local storage
-
Control inference parameters (context size, temperature, top-k, top-p)
-
Maintain conversation context for chat-based workflows
-
No dependency on external services
Ideal for offline deployments, secure environments, or custom fine-tuned models. Developers can build fully self-contained AI systems inside FileMaker.
See Llama functions and Use Llama.cpp in FileMaker.
CoreML
Execute machine learning models using Apple's CoreML framework.
-
Load compiled .mlmodel files
-
Run inference on structured or image data
-
Supports classification, regression, and neural networks
Common use cases include predictive analytics, recommendation systems, and image classification workflows embedded directly in FileMaker.
See CoreML functions and related functions that run CoreML models within the frameworks like CoreImage.Detect function.
Windows ML
Provides access to Windows Machine Learning APIs on Windows 10 and newer.
-
Load ONNX models
-
Perform inference using local hardware acceleration
-
Integrate ML into Windows-based FileMaker deployments
Useful for cross-platform solutions where macOS uses CoreML and Windows uses ONNX-based models.
See WindowsML functions.
Vision
Wraps Apple’s Vision framework for image analysis tasks.
-
Text detection and recognition
-
Face detection and landmark tracking
-
Object recognition and classification
-
Barcode and QR code detection
Often combined with CoreML for custom vision models or with OCR workflows for document processing pipelines.
See Vision functions and Vision Framework for FileMaker.
Image Playground
Interfaces with Apple Intelligence image generation features.
-
Generate images from text prompts
-
Adjust styles and generation parameters
-
Store generated images directly in FileMaker container fields
Useful for generating previews, assets, or user-driven content inside applications.
See ImagePlayground functions and Use Image Playground in FileMaker.
Python
Run Python-based AI workflows directly from FileMaker.
-
Execute Python code via MBS Python functions
-
Call external scripts using Shell functions
-
Integrate libraries such as TensorFlow, PyTorch, scikit-learn, or transformers
This approach enables access to the broader AI ecosystem, including custom pipelines, embeddings, vector databases, and advanced data processing.
See Python functions and Using Python in FileMaker.
OCR
Built-in OCR functionality based on Tesseract.
-
Extract text from images and PDFs
-
Support multiple languages
-
Preprocess images for improved recognition accuracy
Frequently used for invoice processing, document indexing, and data entry automation.
See OCR, WindowsOCR functions and OCR on Windows in FileMaker.
CURL
Provides HTTP client functionality to connect with external AI services.
-
Send REST API requests (POST/GET)
-
Handle authentication (API keys, headers)
-
Parse JSON responses using FileMaker JSON functions
Enables integration with services such as OpenAI, Anthropic, or Google AI models. The plugin includes example files demonstrating ChatGPT API integration and response handling.
See CURL functions and Using ChatGPT in FileMaker.
Integration Patterns
You may combine various functions to build solutions based on the toolbox provided with MBS Plugin:
-
Hybrid AI: Combine local LLMs (FoundationModels/Llama) with cloud APIs
-
Pipeline Processing: OCR → LLM → Structured Data Output
-
AI-assisted UI: Natural language inputs driving FileMaker scripts
-
Offline-first apps: Fully local AI using Llama + CoreML
Conclusion
The MBS FileMaker Plugin enables developers to integrate AI at multiple levels—from low-level ML inference to high-level conversational interfaces.
By combining local processing, external APIs, and scripting bridges, it provides a flexible architecture for building modern, intelligent FileMaker solutions.