Icon

02. GenAI Learnathon_​Deployment_​Solution

<p>GenAI Learnathon: Enrich Data Analytics with GenAI<br><br>During this hands-on learnathon, you will get familiar with the free low-code tool KNIME Analytics Platform, techniques for fraud detection, and the use of GenAI via the KNIME AI Extension.<br><br>Without any coding experience you will learn to:</p><ul><li><p>Detect fraud in investment contracts visually and using statistical techniques</p></li><li><p>Prompt engineer an LLM via the OpenAI integration</p></li><li><p>Create a vector store from a corpus of different types of investment contracts (the knowledge base)</p></li><li><p>Build and deploy a data app to display email alerts and make it available to authenticated users on a web browser</p></li></ul><p>Other KNIME AI integrations we will discuss include GPT4All, Hugging Face, Chroma, FAISS and more.</p>

URL: KNIME AI Extension https://hub.knime.com/knime/extensions/org.knime.python.features.llm/latest/
URL: KNIME for Generative AI https://hub.knime.com/-/collections/~D4ckx2q_J5FPBQXu/

E) Deploy Fraud Detection with GenAI Alert on KNIME Business Hub (OPTIONAL)

Step 1: Apply fraud detection on new contract data

Search, drag and drop a Model Reader and a Numeric Outliers (Apply) to apply the same outlier detection from Part B on new contract data (Part B is included in the workflow 01. GenAI Learnathon_Solution).

Step 2: Apply vector store on newly detected fraud to get context

Search, drag and drop a Model Reader and a Vector Store Retriever to apply the same vector store from Part D on new fraud (Part D is included in the workflow 01. GenAI Learnathon_Solution).

Step 3: Apply prompt engineering and the LLM to perform RAG and generate new fraud email alerts

Copy and paste the String Manipulation and the LLM Prompter from Part D to generate more email alerts and visualize via the provided component.

Step 4: Deploy to KNIME Business Hub the resulting Data App

Upload this workflow and the 'models' folder (located in Relative to > Current workflow) to a KNIME Hub team with an execution context and perform an ad-hoc execution.

Vector store
Model Reader
Prompt engineering
String Manipulation
Outlier detection model
Model Reader
Numeric Outliers (Apply)
Value Lookup
OpenAI Authenticator
DB Table Selector
LLM Prompter
Extract Contract Data
SQLite Connector
OpenAI LLM Connector
PDF Parser
Double-click to configure
Get API Key
DB Reader
Vector Store Retriever
Visualize & downloademail alerts
Email Preview

Nodes

Extensions

Links