From Composer install to your first grounded answer.
-
01
Install
- Composer package (TYPO3 v11, v12, v13, v14)
- Include Static TypoScript template in Root
- Non-Composer mode supported with bundled parsers
composer require t3chatbot/t3-chatbot-ai -
02
Configure
- Chatbot & RAG backend module (per-domain isolation)
- Providers: Local/Remote Ollama, Gemini, OpenAI, Claude, OpenRouter
- Full color palette, custom avatars, and heading titles
- GDPR consent modal for external AI providers
-
03
Automate
- Step 1: Content Discovery (RTE & FAL references)
- Step 2: Vector embedding & indexing with batch limits
- On-the-fly frontend auto-indexing as visitors browse
./vendor/bin/typo3 aichatbot:crawl./vendor/bin/typo3 aichatbot:indexContentClasses --limit=50
-
T3 Chatbot AI is fully tested and compatible with TYPO3 v11.5 LTS, v12.4 LTS, v13.x, and v14.x across PHP 7.4 through PHP 8.4 in both Composer and Non-Composer environments.
-
No. T3chatbot operates non-destructively. It reads published page data and FAL files to build vector representations without altering your page records, Fluid templates, or TCA structures.
-
Yes. AI provider configuration is completely decoupled. You can start with OpenAI or Gemini during development and switch to private on-premise Ollama in production with a single setting in the TYPO3 backend.
-
Yes. T3chatbot indexes permission metadata. Content restricted to specific frontend user groups is excluded from public search queries and only exposed if access control criteria are verified.
-
T3chatbot leverages scalable vector storage adapters (such as Qdrant, Chroma, or database vector extensions). Chunks are indexed asynchronously via CLI commands without degrading frontend web server performance.
-
Yes, absolutely. The TYPO3 backend module provides dedicated appearance settings allowing editors and administrators to configure the primary brand color, heading title, avatar icon, position offsets, and greeting prompts without writing code or touching template files.
