Yesterday Bill Wilder, Udai Ramachandran and I held the first In-person, hands-on, all-day, free. Agent Framework Dev Day in BURLINGTON MA for the year. This year, we targeted both .NET and python developers (last year we only focused on .NET).
My presentation and labs focused on implementing Agentic RAG with Agent Framework.
Talk: Agentic RAG
NOTE: The repository has moved to the new Agent Framework Dev Day github.
About the Lab
This year I tried to stay away from the instructional labs that just turns into a treasure hunt, copy paste set of instructions … though not completely.
This year I created notebooks to cover the fundamentals, then a hands on exercise that had some instructions and excercises - plus the solution to the exercises.
Notebooks:
- The Ingestion Phase (.NET) (python) - fundamentals of document chunking, inserting into Azure AI Search and a comparision of three different embedding models.
- Simple RAG (.NET) (python) - using Azure AI Search to search and interact with LLM to perform RAG
- Advanced RAG (python only) - using the AzureAISearchContextProvider to perform RAG
- Agentic RAG (.NET) (python) - using the Agent Framework to implement a classifier that leads into the hands on excercise.
Hand On Lab:
The starting point was a RAG application that could do semantic search. The end goal was to have an agentic RAG system that could answer 7 of the 9 question types:
- Step 2 (.NET) (pyhton) - walks through implementing an agent to answer the yes/no type questions
- Step 3 (.NET) (pyhton) - walks through implementing an agent to answer the count type questions
- Step 4 (.NET) (pyhton) - provides information on the remaining question types for the attendee to implement themselves
- The solution (.NET) (pyhton) provided the final agentic RAG solution
The sample questions the solution will iterate through if you just run the code:
- “What problems are there with Surface devices?” (Simple question)
- “Are there any issues for Dell XPS laptops?” (Yes/No)
- “How many tickets were logged and Incidents for Human Resources and low priority?” (Count)
- “Do we have more issues with MacBook Air computers or Dell XPS laptops?” (Comparative)
- “Which Dell XPS issue does not mention Windows?” (Difference)
- “What issues are for Dell XPS laptops and the user tried Win + Ctrl + Shift + B?” (Intersection)
- “What department had consultants with Login Issues?” (Multi-hop)
Feedback and Future Workshops
The attendees seemed to be enjoying the day and challenging exercises we had for the day. I was surprised the majority of the attendees were using python and only a handful were using .NET.
We are planning more workshops this year, so if you are interested please reach out to me on twitter @haleyjason (or email - info@jasonhaley.com) and let me know if you are interested in future workshops on agent framework with python or .NET.
If you have a comment, please message me @haleyjason on twitter/X.