Product Overview
Filament RAG adds a production-ready RAG chatbot layer to a Laravel + Filament app.
It is designed for teams that want to ship AI assistants without building the full admin and ingestion stack from scratch.
Best Fit
Filament RAG is a strong fit if you want to:
- add a support or docs bot to an existing Filament product
- launch internal knowledge assistants for staff or admin users
- manage multiple bots with separate sources, prompts, and widget branding
- give non-developers a Filament UI for managing chatbot behavior and content
Probably Not A Fit
It is likely not the right tool if you need:
- a hosted chatbot SaaS with no Laravel or Filament app behind it
- a full billing, tenancy, and customer portal product out of the box
- a generic AI wrapper with no grounding in your own content
In One Sentence
You manage bots, sources, retrieval settings, widget behavior, and conversation history from Filament, while your Laravel app handles runtime, security, and integration.
What This Plugin Adds
- A Filament-native control plane for bots, sources, ingestion, retrieval, and conversations
- An embeddable chat widget for websites, apps, and external frontends
- Queue-driven ingestion for text, files, and public URLs
- Per-bot provider, model, and retrieval settings
- Operational tooling such as setup checks, retries, testing actions, and analytics-friendly workflows
Why Teams Use It
- Add a support, onboarding, or product-help assistant to an existing Filament app
- Launch an internal copilot or staff-facing knowledge assistant
- Run multiple bots with separate prompts, models, sources, and branding
- Let non-developers manage bot behavior and knowledge from inside Filament
What It Does Not Replace
This plugin gives you the chatbot control plane. It does not replace the rest of your application stack.
You still own:
- billing and subscriptions
- tenancy or account isolation
- product-specific workflows
- advanced app permissions beyond the plugin's assistant access controls
- your own source content, policies, and business logic
Why Not Build It Yourself?
You can absolutely build a chatbot stack by hand, but the plugin removes a lot of repeated product work:
- admin CRUD for bots, sources, and conversations
- source ingestion workflows and retries
- retrieval tuning UI and testing actions
- widget embedding, branding, and access controls
- operational checks and privacy workflows
That lets you focus on your product, data, and user experience instead of rebuilding the management layer.
Core Feature Areas
Bot Management
- Multiple bots with separate prompts, models, providers, branding, and answer settings
- Public, member, and admin context areas
- Domain allowlists and signed widget embeds
Knowledge Ingestion
- Text, file, and URL sources
- Queue-based ingestion and retry handling
- Chunking, embedding, and vector persistence
- Re-ingest and rebuild tooling when models or settings change
Retrieval and Answers
- Configurable retrieval depth and similarity filtering
- Source-backed answers with citations
- Streaming responses for widget UX
- Markdown or plain-text answer formats
Widget And Frontend Delivery
- One-script embed for websites
- NPM loader for SPA frameworks
- Titles, subtitles, prompts, welcome text, templates, and accent colors
Operations And Security
- Setup diagnostics through
php artisan filament-rag:doctor - Queue and ingestion visibility in the panel
- Rate limiting, domain restrictions, and signed tokens
- Privacy export and deletion endpoints
Recommended Reading Order
- Core Concepts to understand the product model
- Quickstart to install and validate a clean setup
- Bots to configure assistant behavior
- RAG Sources to add knowledge
- Ingestion and Retrieval to tune answer quality
- Chat Widget to ship the frontend experience