* AI with Aspire *
Aspire with AI
Today I came back across the best AI refactoring video out there done by Steve Sanderson:
https://www.youtube.com/watch?v=yMGTUQhjtlM&ab_channel=dotnet
Code at: https://github.com/dotnet/eShopSupport
UX AI features: chat bots, semantic searches, summarization, type ahead, auto-categorization, etc.
What saves the customer time? Metrics?
Master Detail screen:
semantic searches on search bar = replace former search
summarization of titles column = arrive at title on database save
satisfaction column = sentiment score column
classification column = type column (question, complaint, return)
Child entry screen:
classification dropdown on case type textbox
semantic searches on product textbox
conversation in large multi-line textbox
summarization of conversation in small multi-line textbox
chatbot assistant to search product manuals (RAG) with citations in large multi-line textbox
type ahead in the chatbot
Staff and Customer Apps are different
Maybe add translation of dialog contents to staff, then translate response back.
Need to use Ollama for local testing at minimum:
https://ollama.com/
Microsoft.Extensions.AI.Abstractions assembly dll
Microsoft.Extensions.AI library
IChatClient interface
ChatClientBuilder Class
UseOpenTelemetry(ChatClientBuilder, ILoggerFactory, String, Action<OpenTelemetryChatClient>) extension method
IImageGenerator interface
IEmbeddingGenerator interface
Semantic search is an information retrieval technique that understands the context, intent, and meaning behind a user's query, rather than just matching keywords.
Implementing semantic search in C# typically involves three core steps: 1) converting text into numerical embeddings using an AI model, 2) storing these in a vector database (such as Elasticsearch or Azure AI Search), and 3) performing a similarity search.
Elasticsearch Client for Semantic Searches:
https://www.elastic.co/elasticsearch/vector-database
Jovanovic's
https://www.youtube.com/watch?v=4B3ppx2U8bE
Comments
Post a Comment