Did you miss Tech Talk Live #168 - AI-Powered Content Enrichment and Transformation Workflows for Alfresco?
Angel Borroy introduced an open-source pipeline for Knowledge Enrichment:
1. Markdown Transformer: turns Office/PDF files into Markdown, then calls the vision-capable LLaVA model to add image descriptions.
2. Embedding Generator: feeds that Markdown into Elasticsearch via Spring AI, giving you semantic search without Hyland CIC.
Daniel Gradecak presented an alternative approach to AI and Alfresco written in Spring Boot + LangChain (Python)
1. Summariser: fills an Alfresco property (like "cm:description") using the BART-Large-CNN model
2. Categoriser: a fast TF-IDF + Naïve Bayes engine you train via REST to tag documents with your own metadata
3. Prometheus Metrics Exporter: pulls Alfresco Search stats and exposes them for Prometheus
4. Command-Line Transformer Template: drop-in Docker image so any shell tool (e.g., ExifTool, ImageMagick) can become an Alfresco transformer
Take-home insight (my pick): Unless you need Hyland CIC, a lean Docker stack (Markdown → embeddings → Elasticsearch) plus a couple of custom T-Engines already gives you state-of-the-art AI enrichment, while keeping data on-prem and costs predictable.
Community news: Early-bird for CommunityLIVE (Las Vegas) ends in June; Alfresco founder John Newton is back at Hyland and will speak at CommunityLIVE 2025.
The recording is ready, please help yourself:
You will also find the slides and the transcript of the video attached at the bottom of this article.
Next TTL will be in June, stay tuned.
