Building Your First Multi-Modal Knowledge Assistant
As organizations work to develop interactive, intelligent GenAI Assistants, they quickly realize that the majority of their data exists in unstructured or semi structured documents, PowerPoints, or emails, not in structured databases. This is the data that answers your users' questions of 'Why' and 'So What'...
The skills required to develop interactive GenAI assistants that leverage multi-modal data sources are fundamentally new to most developers, and require focus to master. This is particularly true for developers new to working with non-deterministic GenAI capabilities that require equal parts development and data science mindsets.
Workshop is a guided, hands-on learning experience where your team builds a multi-modal RAG pipeline step-by-step. Upon completion, your team won’t just understand how to work with unstructured PDFs, tables, and charts – they’ll have built a working multi-modal assistant with dev practices they can apply in their daily role.
- Working with Complex Text - Extract and prepare textual content from complex PDF sources.
- Working with Rich Tabular Data - Identify, parse, and structure tabular data for retrieval.
- Working with Embedded Charts, Graphs, & Images - Process charts and graphs to generate usable representations.
- Creating a Full-Functioning Multi-Modal Assistant - Build a full-stack knowledge assistant in a full IDE.
- Confidence Working with Unstructured Data - Learn practical techniques for handling text, images, and tables.
- A Ready-to-Use Multi-Modal Pipeline - Leave with a working end-to-end assistant.
- Higher Retrieval Accuracy - Improve grounding by combining signals across formats.
- Reusable Multi-Modal Patterns - Apply consistent ingestion and retrieval approaches to future projects.
- Greater Readiness for Enterprise RAG Solutions - Equip your team to build real, production-minded assistants.
Who Should Attend:
Solution Essentials
Virtual or in-person
4 hours
Basic Python recommended
Jupyter notebooks plus preconfigured multi-modal RAG components