Elastic is in an unusual position in the current AI cycle. Retrieval is the foundation of every RAG system anyone is building, Elastic has been doing search seriously for well over a decade, and that depth shows in the training material in a way that newer vector database vendors cannot match. The free on demand modules now include RAG foundation, semantic search foundation, text embedding for semantic search, an introduction to MCP with the Elasticsearch MCP server, and agent builder material covering tools and agents. Making all of that free, along with the associate accreditation badges, is a meaningful decision and those modules are the strongest reason to engage with this programme.
What I particularly value is that this material teaches retrieval as search rather than as embeddings. A great deal of RAG education proceeds as though the whole problem were choosing an embedding model and a vector store, and the actual determinants of quality are chunking strategy, hybrid retrieval combining lexical and semantic signals, ranking, filtering and relevance tuning. Those are old search problems and the people teaching here know them properly. Anyone whose RAG system returns confidently irrelevant results will learn more from the semantic search foundation module than from most dedicated AI courses.
The Jina AI acquisition in October 2025 brought embeddings and reranking expertise in house, and the material has benefited. The paid side is where I have reservations, and they are about access rather than quality. Certification exams run somewhere between four and five hundred dollars per attempt depending on the track, and published figures vary enough between sources that you should check before booking. There are no discounted retakes, a fourteen day wait after a failure, non refundable fees and a year to use them.
The exams are hands on and genuinely difficult, which is exactly what makes them worth something in the job market, and the pricing plainly assumes an employer is paying. An individual switching careers is looking at a serious outlay with a real chance of having to pay twice. Instructor led course and training subscription pricing is not published at all, which is standard enterprise practice and unhelpful for anyone trying to plan. The Certified Engineer exam moves from 8.15 to 9.3 on the first of September 2026, so timing matters if you are preparing.
The other irritation is naming. Elastic has moved through Elasticsearch Relevance Engine and now Search AI and Agent Builder positioning in a fairly short span, which leaves older material and community writing referring to concepts under names that have since changed. The underlying technology is more stable than the branding suggests, and a newcomer cannot easily tell. Three point six.
Excellent, genuinely current free material on the retrieval fundamentals that most AI education skips, marked down for exam pricing aimed squarely at corporate budgets and for branding that keeps moving.