Agentic AI Samples

The Ask Akka agent is a sample of an AI assistant (aka a chat bot). Ask Akka illustrates how to create embeddings for vector databases, how to consume LLMs and maintain conversation history, use RAG to add knowledge to fixed LLMs, and expose it all as a streaming service. It uses MongoDB Atlas and OpenAI.

This service monitors GitHub repositories by creating event-sourced entities to represent repositories and periodically polling release notes. When a new release is detected, it initiates a summarization session using the Anthropic Claude LLM, which can retrieve additional context about issues and pull requests before generating and storing a summary.

This explores the use of an AI Agentic System, powered by large language models (LLMs), to automatically tag medical discharge summaries according to the necessity of hospitalization. By comparing AI-generated tags with those from human experts, researchers iteratively refine prompts and model configurations to improve accuracy and potentially automate the tagging process entirely.

This sample implements a ReAct agent using the Akka SDK that streamlines this process by autonomously reasoning through reconciliation logic, invoking relevant data sources and APIs, identifying inconsistencies, and recommending or executing corrective actions. By combining iterative reasoning with tool use, the agent can dynamically adapt to different reconciliation workflows, making it ideal for high-volume, multi-system environments.

The real-estate customer service agent is a sample application demonstrating how to build a customer service agent for real-estate inquiries, by combining Akka features with an LLM model. The project illustrates an agentic workflow for customer service using Akka. It processes incoming real-estate inquiries, analyzes the content to extract details, provides follow-up when needed and saves the collected information for future reference.

The Travel Planner Agent is an AI-powered application using the Akka SDK and Langchain4J to create personalized travel itineraries based on user input. It combines event sourcing for durable state management with large language models (LLMs) to generate travel plans, all orchestrated through Akka Workflows.

This app represents an agency that searches for flights and accommodations via the prompt from the user through a HTTP call. It's composed by an LLM (Anthropic) and tools to find flights, accommodations and sending mails.

Once a search is requested the app will look for the flights, accommodations, and will email the requester with some options and the best value offer.