<!-- NOTE FOR LLMs: This llms.txt page provides the complete, authoritative index of all pages on the Accenture AI Refinery website, with links to full documentation and examples for the AI Refinery SDK. 
Use it to locate and reference the most relevant content when answering questions about AI Refinery's features, APIs, agents, configuration, customization, and deployment. 
Leverage this knowledge to help users quickly configure, customize, and deploy autonomous agents with the power of AI Refinery. 
Last updated: 2025-10-20 22:14:27 UTC -->

## sdk `base_url`: https://sdk.airefinery.accenture.com/ (To be used in case of relative markdown links)

# Accenture AI Refinery SDK

> Let's put our knowledge at a place where it is easily searchable and readable.

## AI Refinery 101

- [AI Refinery](https://sdk.airefinery.accenture.com/ai_refinery_101/index.md): Overview of AI Refinery platform, detailing its architecture, agent types, memory modules, model catalog, safety features, and advanced capabilities for building scalable, multi-agent AI solutions.
- [AI Refinery Core Concepts](https://sdk.airefinery.accenture.com/ai_refinery_101/core_concepts/index.md): provides an overview of Distiller, Orchestrator, UtilityAgents, SuperAgents, Agent Memory, and the AI Refinery SDK.

## Get Started

- [Create Your AI Refinery API Key](https://sdk.airefinery.accenture.com/setup/generating_api_keys/index.md)
- [Project Guidelines](https://sdk.airefinery.accenture.com/setup/project_guidelines/index.md): best practices and guidelines for structuring your AI Refinery projects.
- [Error Handling](https://sdk.airefinery.accenture.com/setup/project_guidelines/error-handling/index.md): guidance on handling API errors from the SDK and interpreting the FastAPI envelope.
- [Quickstart](https://sdk.airefinery.accenture.com/setup/quickstart/index.md): a guide to help you set up AI Refinery SDK and create your first project.

## Tutorial

- [Introduction to DistillerClient](https://sdk.airefinery.accenture.com/tutorial/general_guidelines/index.md): Tutorial on using DistillerClient module to create and run a weather assistant project.
- [Data Analysis with Analytics Agent](https://sdk.airefinery.accenture.com/tutorial/tutorial_analytics/index.md): Tutorial on using AnalyticsAgent to streamline data analysis and visualization of different information from PostgreSQL databases or CSV files.
- [Prompt Compression and Reranking](https://sdk.airefinery.accenture.com/tutorial/tutorial_compression_reranker_api/index.md): Tutorial on integrating prompt compression PAI and reranking API.
- [Evaluation Super Agent - agent performance assessment system](https://sdk.airefinery.accenture.com/tutorial/tutorial_evaluation_superagent/index.md): Tutorial on using the EvaluationSuperAgent to assess UtilityAgent performance.
- [Employing Built-in Agents](https://sdk.airefinery.accenture.com/tutorial/tutorial_garden/index.md): Tutorial demonstrating how to create a gardening-focused AI assistant using a built-in UtilityAgent in AI Refinery.
- [Human Agent - Human in the Loop](https://sdk.airefinery.accenture.com/tutorial/tutorial_human/index.md)
- [Agent Memory - Memory Storage and Retrieval](https://sdk.airefinery.accenture.com/tutorial/tutorial_memory/index.md): Tutorial on integrating memory modules to enable context retention and personalized responses.
- [Handling Complex Tasks with Super Agents](https://sdk.airefinery.accenture.com/tutorial/tutorial_myth_buster/index.md): Tutorial on coordinating workflows between UtilityAgents and a SuperAgent to solve complex tasks.
- [Orchestrator - Customization](https://sdk.airefinery.accenture.com/tutorial/tutorial_orchestrator/index.md)
- [RAG with Semantic/Vector Search](https://sdk.airefinery.accenture.com/tutorial/tutorial_park_information/index.md): Tutorial on building a Retrieval Augmented Generation (RAG) system with semantic/vector search by using ResearchAgent.
- [Combine Your Custom Agents with Built-in Agents](https://sdk.airefinery.accenture.com/tutorial/tutorial_party/index.md): Tutorial on combining CustomAgents with built-in UtilityAgents to create a party planning assistant.
- [PII (Personally Identifiable Information) Masking Module](https://sdk.airefinery.accenture.com/tutorial/tutorial_pii_masking_module/index.md): Tutorial on the PII Masking Module, built on Microsoft Presidio, to securely detect and mask personally identifiable information (PII).
- [Responsible AI Module - safety and policy checks to safeguard queries](https://sdk.airefinery.accenture.com/tutorial/tutorial_rai_module/index.md): Tutorial on using the Responsible AI (RAI) Module to enforce safety and ethical standards for LLM queries.
- [Self-reflection - Automatically Refine Agent Response](https://sdk.airefinery.accenture.com/tutorial/tutorial_self_reflection/index.md): Tutorial on enabling Self-Reflection in UtilityAgents.
- [Retrieval Augmented Generation (RAG) with Web Search](https://sdk.airefinery.accenture.com/tutorial/tutorial_shopping_helper/index.md): Tutorial on building a Retrieval Augmented Generation (RAG) with Web Search by using SearchAgent and CustomAgent.
- [Enable Tool Usage by Agents](https://sdk.airefinery.accenture.com/tutorial/tutorial_tool_use/index.md): Tutorial on how to integrate custom Python functions with the ToolUseAgent.
- [Utilize A2A-exposed Agents](https://sdk.airefinery.accenture.com/tutorial/a2a_agents/a2a_currency_converter/index.md): Tutorial on using A2AClientAgent with A2A servers.
- [Employ Databricks Agents](https://sdk.airefinery.accenture.com/tutorial/databricks_database_assistant/database_assistant/index.md): Tutorial on using Databricks Agent that allow user to talk with their data stored in a Databricks.
- [Flow Super Agent - A Super Agent with Deterministic Workflow](https://sdk.airefinery.accenture.com/tutorial/flow_superagent/tutorial_flow_superagent/index.md): Tutorial provides an example of using FlowSuperAgent to manage other UtilityAgent(s).
- [Triage for Flow Super Agent - Deterministic Workflows with Conditional Execution](https://sdk.airefinery.accenture.com/tutorial/flow_superagent/tutorial_flow_superagent_triage/index.md)
- [Generating Images - Text-to-Image and Image-to-Image](https://sdk.airefinery.accenture.com/tutorial/img_generation/tutorial_img_generation/index.md): Tutorial on building the ImageGenerationAgent by using UtilityAgent (Report Agent), SearchAgent (Search Agent), ImageUnderstandingAgent (Image Understanding Agent), ImageGenerationAgent (Image Generation Agent), and UtilityAgent (Story Teller Agent).
- [Understanding Images - Chart Reading, OCR & more](https://sdk.airefinery.accenture.com/tutorial/img_understanding/tutorial_img_understanding/index.md): Tutorial on building an ImageUnderstandingAgent by using SearchAgent (Search Agent), ImageUnderstandingAgent (Image Understanding Agent), UtilityAgent (Story Teller Agent) and ImageUnderstandingAgent (Markdown Agent).
- [Creating & Utilizing a Vector Search Index](https://sdk.airefinery.accenture.com/tutorial/knowledge_extraction/knowledge_extraction/index.md): Tutorial on building the Knowledge Extraction Agent by using CustomAgent and ResearchAgent.
- [Knowledge Graph/GraphRAG Semantic Layer](https://sdk.airefinery.accenture.com/tutorial/knowledge_graph/knowledge_graph/index.md): Tutorial on how to use knowledge Graph API to create, update, query, and visualize entity-relation graphs extracted from raw documents.
- [Employ Snowflake Agents](https://sdk.airefinery.accenture.com/tutorial/snowflake_agent/snowflake_walkthrough/index.md)

## Agent Library - Overview

- [Overview](https://sdk.airefinery.accenture.com/distiller/agent-library/index.md): Introduce the Agent Library, a resource for discovering and integrating deployable agents to enhance data processing and AI-powered projects.

## Agent Library - Utility

- [Utility Agents](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/index.md): Introduction of UtilityAgents - designed to streamline tasks like RAG, analytics, and image processing, enhancing workflow efficiency and project execution.
- [Analytics Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/analytics/index.md): Overview of the AnalyticsAgent, which provides advanced data analysis capabilities, connecting to data sources like PostgreSQL to retrieve SQL queries, perform complex analytics, and help users extract valuable insights within their projects.
- [Author Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/author/index.md): Overview of the AuthorAgent, which assists users in generating and refining content, providing a streamlined approach to content creation and editing within AI Refinery projects.
- [Critical Thinker Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/criticalthinker/index.md)
- [Human Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/human/index.md)
- [Image Generation Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/imagegeneration/index.md)
- [Image Understanding Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/imageunderstanding/index.md)
- [Planning Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/planning/index.md): The PlanningAgent assists in task management and project planning, helping users organize and prioritize tasks effectively within their AI Refinery projects.
- [Search Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/search/index.md): The SearchAgent provides advanced search capabilities, a subclass of ResearchAgent, answers user queries by transforming them for Google Search, retrieving relevant results, and generating detailed responses.
- [Tool Use Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/tool_use/index.md): The ToolUseAgent enables agents to utilize external tools and APIs, executes user-requested tasks by selecting and running predefined external tools, then returning the results seamlessly.
- [A2A Client Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/A2A Client Agent/index.md): Overview of the A2AClientAgent, which allows users to interact with A2A servers, enabling seamless integration and communication between agents.
- [A2A Server Catalog](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/A2A Client Agent/a2a_catalog_overview/index.md): Overview of the A2AClientAgent's catalog, showcasing two implementations of A2A servers that are ready for use both locally and in the AI Refinery platform.
- [A2A Server Introduction](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/A2A Client Agent/a2a_intro/index.md): Introduction to the A2AClientAgent, which facilitates communication with A2A servers, enabling agents to access and utilize external resources and services.
- [Deep Research Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/Deep Research Agent/deep_research/index.md)
- [MCP Client Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/MCP Client Agent/index.md): Overview of the MCPClientAgent, which allows users to interact with MCP servers, enabling seamless integration and communication between agents.
- [MCP Server Catalog](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/MCP Client Agent/mcp_catalog_overview/index.md): The MCPClientAgent's catalog showcases ready-to-use MCP server implementations, available locally and in the AI Refinery platform, highlighting varied setup requirements and the broad capabilities they offer to LLMs.
- [MCP Server Introduction](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/MCP Client Agent/mcp_intro/index.md): Introduction to the MCPClientAgent, which facilitates communication with MCP servers, enabling agents to access and utilize external resources and services.
- [Research Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/Research Agent/index.md): Overview of the ResearchAgent, which provides advanced research capabilities using one or more Retrievers from the retrievers_gallery below.
- [Retrievers Gallery](https://sdk.airefinery.accenture.com/distiller/agent-library/utility_agents/Research Agent/retrievers_gallery/index.md): A gallery of retrievers available for the ResearchAgent, such as WebSearchRetriever, AzureAISearchRetriever, ElasticSearchRetriever, and customizable options—that fetch relevant information from diverse sources, enhancing its research capabilities.

## Agent Library - Super Agents

- [Super Agents](https://sdk.airefinery.accenture.com/distiller/agent-library/super_agents/index.md): Overview of SuperAgents, which are advanced agents designed to handle complex tasks and workflows, integrating multiple UtilityAgents to provide comprehensive solutions.
- [Evaluation Super Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/super_agents/evaluation_superagent/index.md): The EvaluationSuperAgent uses predefined metrics and sample or generated queries, providing a structured process for measuring and improving agent performance.
- [Flow Super Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/super_agents/flow_superagent/index.md): The FlowSuperAgent is a deterministic SuperAgent that executes a predefined workflow, ensuring consistent and reliable task completion by following a structured sequence of operations.
- [Super Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/super_agents/super_agent/index.md): The SuperAgent manages complex, multi-step tasks by breaking them into subtasks, assigning them to appropriate UtilityAgents, and iteratively completing them using a structured checklist until a final, comprehensive response is produced.

## Agent Library - User Defined Agents

- [Base Utility Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/user_defined_agents/baseutilityagent/index.md): The UtilityAgent is a lightweight, customizable agent for simple, single-task prompts—like summarization or formatting—guided by a magic_prompt, with flexible output formatting and optional context inclusion.
- [Custom Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/user_defined_agents/customagent/index.md): The CustomAgent lets you define custom Python logic for anything from simple queries to complex workflows, executing locally and integrating into the platform via an executor_dict for orchestration.

## Agent Library - Trusted Agent Huddle

- [Trusted Agent Huddle](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/index.md): The Trusted Agent Huddle securely integrates third-party and built-in agents, streamlining workflows and enabling an agent trust score for greater efficiency.
- [Amazon Bedrock Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/awsagent/index.md): Amazon Bedrock Agent, a third-party agent that provides access to various AI models and services on the Amazon Bedrock platform, enabling users to leverage advanced AI capabilities within their projects.
- [Azure AI Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/azureaiagent/index.md): Azure AI Agent, a third-party agent that integrates with Azure's AI services, allowing users to access and utilize Azure's powerful AI capabilities within their projects.
- [CB Insights Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/cb_insights/index.md)
- [Databricks Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/databricks/index.md): Databricks Agent, a third-party agent that connects to Databricks' data analytics platform, uses Genie for natural language data interaction and integrates with the AI Refinery SDK via the DatabricksAgent class to enhance capabilities.
- [Google Vertex Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/google_vertex/index.md)
- [Pega Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/pega/index.md)
- [Salesforce Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/salesforce/index.md): Salesforce Agent, a third-party agent that integrates with Salesforce's AI services, handles support, data, and scheduling tasks, integrating with the AI Refinery SDK via the SalesforceAgent class to extend capabilities.
- [SAP Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/sapagent/index.md)
- [ServiceNow Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/servicenow/index.md)
- [Snowflake Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/snowflake/index.md)
- [Wolfram Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/wolfram/index.md)
- [Writer AI Agent](https://sdk.airefinery.accenture.com/distiller/agent-library/third_party_agents/writer/index.md): Writer AI Agent, a third-party agent that integrates with Writer's AI services, providing advanced content generation and editing capabilities within the AI Refinery SDK.

## Model Catalog

- [Overview](https://sdk.airefinery.accenture.com/distiller/model_catalog/index.md)
- [AI-Transcription](https://sdk.airefinery.accenture.com/distiller/model_catalog/ASR/Azure/AI-Transcription/index.md)
- [llmlingua-2-bert-base-multilingual-cased-meetingbank](https://sdk.airefinery.accenture.com/distiller/model_catalog/Compression/microsoft/llmlingua-2-bert-base-multilingual-cased-meetingbank/index.md)
- [FLUX.1-schnell](https://sdk.airefinery.accenture.com/distiller/model_catalog/Diffusers/black-forest-labs/FLUX.1-schnell/index.md)
- [Qwen3-Embedding-0.6B](https://sdk.airefinery.accenture.com/distiller/model_catalog/Embedding/Qwen/Qwen3-Embedding-0.6B/index.md)
- [e5-mistral-7b-instruct](https://sdk.airefinery.accenture.com/distiller/model_catalog/Embedding/infloat/e5-mistral-7b-instruct/index.md)
- [multilingual-e5-large](https://sdk.airefinery.accenture.com/distiller/model_catalog/Embedding/infloat/multilingual-e5-large/index.md)
- [llama-3-2-nv-embedqa-1b-v2](https://sdk.airefinery.accenture.com/distiller/model_catalog/Embedding/nvidia/llama-3-2-nv-embedqa-1b-v2/index.md)
- [nv-embedqa-mistral-7b-v2](https://sdk.airefinery.accenture.com/distiller/model_catalog/Embedding/nvidia/nv-embedqa-mistral-7b-v2/index.md)
- [Qwen3-32B](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/Qwen/Qwen3-32B/index.md)
- [Deepseek-r1-distill-qwen-32b](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/deepseek-ai/deepseek-r1-distill-qwen-32b/index.md)
- [Llama-3.1-70B-Instruct](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/meta-llama/llama-3.1-70B-Instruct/index.md)
- [Llama-3.1-8B-Instruct](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/meta-llama/llama-3.1-8b-instruct/index.md)
- [Llama-3.2-90B-Vision-Instruct](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/meta-llama/llama-3.2-90B-Vision-Instruct/index.md)
- [Llama-3.3-70b-Instruct](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/meta-llama/llama-3.3-70b-instruct/index.md)
- [Mistral-7B-Instruct-v0.3](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/mistralai/Mistral-7B-Instruct-v0.3/index.md)
- [Mistral-Small-3.1-24B-Instruct-2503](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/mistralai/Mistral-Small-3.1-24B-Instruct-2503/index.md)
- [gpt-oss-120b](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/openai/gpt-oss-120b/index.md)
- [gpt-oss-20b](https://sdk.airefinery.accenture.com/distiller/model_catalog/LLMs/openai/gpt-oss-20b/index.md)
- [bge-reranker-large](https://sdk.airefinery.accenture.com/distiller/model_catalog/Reranker/BAAI/bge-reranker-large/index.md)
- [llama-3.2-nv-rerankqa-1b-v2](https://sdk.airefinery.accenture.com/distiller/model_catalog/Reranker/nvidia/llama-3.2-nv-rerankqa-1b-v2/index.md)
- [sam-hq-vit-base](https://sdk.airefinery.accenture.com/distiller/model_catalog/Segmentation/syscv-community/sam-hq-vit-base/index.md)
- [AI-Speech](https://sdk.airefinery.accenture.com/distiller/model_catalog/TTS/Azure/AI-Speech/index.md)

## API

- [API Reference](https://sdk.airefinery.accenture.com/api-reference/index.md): An API reference guide for the AI Refinery SDK.
- [Chat Completion](https://sdk.airefinery.accenture.com/api-reference/chat-completions-index/index.md): Chat Completion API generates dynamic and contextually appropriate responses using advanced language models from the model catalog.
- [Distiller](https://sdk.airefinery.accenture.com/api-reference/distiller-index/index.md): Use Distiller API to build collaborative multi-agent systems with advanced reasoning capabilities.
- [Embeddings](https://sdk.airefinery.accenture.com/api-reference/embeddings-index/index.md): Use Embeddings API to generate vector representations of text for tasks like semantic search and clustering.
- [Models](https://sdk.airefinery.accenture.com/api-reference/model-index/index.md): Use Models API to retrieve a list of available models for different tasks.
- [Moderations](https://sdk.airefinery.accenture.com/api-reference/moderations-index/index.md)
- [Physical AI (preview)](https://sdk.airefinery.accenture.com/api-reference/physical-AI-index/index.md): Physical AI is currently in preview and not yet avaiable for use. The PhsyicalAI API provides advanced tools for vidoe-based understanding, simulation, and sysnthesis of the physical world.
- [Automatic Speech Recognition (ASR)](https://sdk.airefinery.accenture.com/api-reference/audio_api/asr-transcription-index/index.md): Using Automatic Speech Recognition (ASR) Transcription API convert audio files into text.
- [Text-to-Speech (TTS)](https://sdk.airefinery.accenture.com/api-reference/audio_api/tts-speech-index/index.md): Using Text-to-Speech (TTS) API to convert text into audio.
- [Image Generation](https://sdk.airefinery.accenture.com/api-reference/image_api/img-gen-index/index.md): Use Image Generation API create images from text prompts.
- [Image Segmentation](https://sdk.airefinery.accenture.com/api-reference/image_api/img-seg-index/index.md): Use Image Segmentation API to identify and label regions within an image based on user-provided point prompts.
- [Knowledge API Reference](https://sdk.airefinery.accenture.com/api-reference/knowledge_api/knowledge-api-index/index.md)
- [Knowledge Extraction](https://sdk.airefinery.accenture.com/api-reference/knowledge_api/knowledge-extraction-index/index.md): Use Knowledge Extraction API to process documents and extract structured information for downstream search and retrieval tasks.
- [Knowledge Graph](https://sdk.airefinery.accenture.com/api-reference/knowledge_api/knowledge-graph-index/index.md): Use Knowledge Graph API to build, update, visualize, and query knowledge graphs from extracted document content.
- [Fine-Tuning](https://sdk.airefinery.accenture.com/api-reference/training_api/fine-tuning-index/index.md)

## Release Log


## Support

- [Support](https://sdk.airefinery.accenture.com/support/index.md): Support page for AI Refinery SDK, providing resources and contact information