Interface
See it in action
Purpose-built for FileMaker Pro — every screen designed for speed and clarity.
AI Playground
Prompt management & history
Prompt library, multi-model router, and evaluation suite — using only native FileMaker 2025 script steps. No plugins. No addons.
Everything a FileMaker developer needs to run serious AI workflows — without ever leaving their solution.
Route any prompt to any model — GPT-4o, Claude, Mistral, Llama — via OpenRouter, LM Studio, or any OpenAI-compatible endpoint. Endpoints, models, and sources are fully dynamic at runtime.
Designate a judge model to score and reason about output quality automatically. Evaluation as a first-class workflow.
Author and version prompts in JSON, Markdown, or plain text — with database variable injection baked in.
Every run logged in full: model, endpoint, temperature, prompt version, output. Complete observability.
Integrates with the Agentic FileMaker Coding framework for multi-step AI workflows.
Version 0.2 turns fmRouter from a prompt runner into a proper evaluation and observability platform.
Prompts can now be Markdown, plain text, or any format. Opens fmRouter to creative, instructional, and complex tool-calling workflows.
Attach rubrics to any prompt. Route outputs to a judge model for automated scoring. Compare results across runs, models, and versions.
Connect prompt and evaluation infrastructure to agentic-fm for multi-step AI coding workflows directly in FileMaker.
Purpose-built for FileMaker Pro — every screen designed for speed and clarity.
Clone the repo, explore the schema, or request a live demo file.