Skip to main content
Version: v5.0

AI Service overview

Overview#

The AI Service introduces a range of advanced AI features to the platform. This new service is introduced in Platform version 5.0 and offers a collection of AI components which facilitate a more natural human-language type interaction with platform data.

The AI Service enables application developers to define and configure intelligent agents which can process user prompts, retrieve real-time data and return intelligent, context-aware responses.

AI Service features#

The AI Service introduces the following features:

  • Users can seamlessly execute AI agents.
  • Several agents can collaborate to handle tasks in multi-agent workflows. ​
  • Agents can auto-select the most appropriate tools for each task. ​
  • AI Teams group multiple agents in a team to interact sequentially to achieve a goal.
  • Project administrators can securely configure Large Language Model (LLM) API keys for the AI Service to use.​
  • You can assign preferred LLMs to each agent. Currently, the platform supports all OpenAI models.​
  • Deploy your own custom agents and tools​

Note: For a good understanding of basic AI concepts refer to the AI Concepts page.

AI Service resources#

The AI Service provides the following first-class resources that are owned by a namespace. Users can create these resources and share them with other users with the platform permissions model.

Note: All platform users can access tool and agent resources.

The relationship between AI service resources is as follows:

  • A conversation recruits a team
  • A team manages agents
  • Agents use tools and reference knowledgebases

For more information, see the following AI Service resources:

  • Agent: An AI-powered component that understands input, can invoke tools, and then responds based on its assigned role. You can configure agents to process user prompts and retrieve real-time data.
  • Team: A group of agents configured to work together sequentially to solve a task.
  • Tool: A utility or API integration that helps agents retrieve or process platform data.​ It can be either a custom tool you create or a system tool.
  • KnowledgeBase: A collection of helpful information that the LLM can search through in order to answer questions more accurately​.
  • Conversation: The interactive session that maintains context between the user and the team across multiple turns.

Sample data flow between Agents, Teams, and APIs#

Figure: Data flow within AI Service

AI Service API key#

Project administrators must store the OPENAI_API_KEY to enable the AI Service. The OPENAI_API_KEY is maintained separately for each namespace.

The key is encrypted both at test and in transit. Trusted clients such as the AI Service can securely retrieve the decrypted secret.