Jun 17, 2026

RAG's Not Dead, You're Just Using It Wrong! - Phil Nash

PDFs were designed for display, not data extraction. This is a primary reason why your RAG system might be failing.

#1about 1 minute

Why retrieval-augmented generation is misunderstood, not obsolete

RAG is often dismissed as a solved problem, but its effectiveness depends on customizing it for specific data, users, and use cases.

#2about 2 minutes

Overcoming the challenge of extracting data from PDFs

The PDF format was designed for display, not data extraction, which complicates the initial data ingestion step in a RAG pipeline.

#3about 1 minute

An overview of the OpenRAG open source framework

OpenRAG provides a customizable RAG baseline by combining open source tools like Dockling for parsing, OpenSearch for search, and Langflow for orchestration.

#4about 1 minute

How agentic RAG improves on traditional approaches

Agentic RAG empowers the model to decide when to query a database, ask follow-up questions, and use other tools, making it more powerful than a simple retrieval flow.

#5about 3 minutes

Using RAG to access private and post-training data

RAG is essential for querying internal, private documents and accessing information created after a model's training cutoff date without compromising data security.

#6about 3 minutes

How IBM's coding agent modernizes legacy enterprise code

IBM's coding agent, Bob, is designed to help large enterprises modernize legacy codebases like COBOL and Java by intelligently selecting the right model for each task.

#7about 4 minutes

Strategies for developers to manage AI information overload

To keep up with the fast pace of AI, developers can rely on curated newsletters to filter out hype and focus on verified, useful information.

#8about 4 minutes

Integrating OpenRAG with Twilio for voice AI applications

OpenRAG can serve as the knowledge and action layer for a voice AI agent built with Twilio's ConversationRelay or Agent Connect, handling text-based interactions over the phone.

#9about 2 minutes

Exploring business models for open source AI tools

Open source projects like OpenRAG can be monetized through enterprise versions that offer specific features, such as limiting capabilities for compliance, rather than adding them.

#10about 6 minutes

The nuanced impact of AI on developer productivity

AI coding tools can accelerate work for those learning a new area but may slow down experts who must spend time verifying and debugging the generated code.