Creatio AI architecture
This article covers the overall structure and architecture of Creatio AI.
Creatio AI is the AI architecture at the core of the Creatio platform, designed to empower organizations with intelligent automation by acting as a virtual assistant to the end user performing tasks. All Creatio composable applications, including both the modern CRM apps for marketing, sales and service as well as and custom no-code apps, can leverage Creatio AI for advanced AI efficiencies and intelligent workflow.
Creatio AI's unified AI approach brings together prescriptive, generative, and agentic AI to deliver a comprehensive solution for modern business automation.
This article discusses the lower level components of the Creatio AI architecture, to give a better understanding of the major technologies and components.
Creatio AI engine
The Creatio AI engine is the main component of Creatio AI. Its primary responsibilities include the following:
- Initiate the execution of actions.
- Make requests to the Creatio platform.
- Determine the next steps in the workflow.
- Dispatch messages to microservices.
- Ensure consistency and efficiency in system operations.
This component oversees the complete cycle of processing user requests, guaranteeing that all actions are executed properly and in the correct order.
Context
The context refers to any additional information that Creatio AI can use to formulate responses, launch actions, and more.
Creatio AI supports reading data only from Freedom UI pages and retrieves information only up to 30 records from a list page.
Out of the box, the context of Creatio AI consists of the following main elements:
-
Information from the user's currently open page. Creatio AI works with page information on demand. Out of the box, the system only receives the following:
- the data structure of the page
- the identifier ID of the open record
- the primary column for display
To obtain any additional information, Creatio AI makes separate requests to Creatio.
-
Chat message history. Creatio AI has access to the entire message history in the current chat. It does not have access to the history of previous chats or chats of other users.
-
Results of executing actions. While operating, Creatio AI might execute AI Skills and actions to achieve the user's goal. In such cases, it has access to parameters marked as output parameters in the actions. This lets you pass any additional information to Creatio AI and adjust its behavior based on the results of action execution. Creatio AI can also execute system actions. Learn more in a separate article: System actions.
Creatio AI Skills
Creatio AI Skills are individual business tasks or scenarios that Creatio AI can perform upon user request. Each skill consists of the following:
- a prompt for the Large Language Model (LLM)
- a list of actions that Creatio AI can execute for the user
This lets you extend Creatio AI functionality by adding new AI Skills tailored to specific business needs. Learn more in a separate article: Develop Creatio AI Skill.
System skill
A System akill is a special entity Creatio AI uses to ensure correct behavior and functionality when interacting with the user. The system skill defines the following:
- how interactions with the user in the chat must occur
- how links are formed
- what the scope of responsibility for Creatio AI is
- what the rules for handling basic requests to Creatio AI or actions are
A key component of the system skill are system actions. Learn more in a separate article: System actions.
Custom actions
Custom actions are user-made functionality that Creatio AI can perform to achieve the user's goal described in an AI Skill. For example, using custom actions, Creatio AI can perform agentic actions such as the following:
- Perform operations in Creatio.
- Retrieve new data.
- Open pages.
- Modify information.
- Invoke web services and ML models.
You can create and customize custom actions using an intuitive visual designer, ensuring a straightforward development process without the need for programming.
Microservice
Creatio AI uses a microservice architecture to interact with the LLM. The responsibilities of the microservice include the following:
- Rout requests to the LLM model.
- Handle errors.
- Check the current limits of the client.
This approach provides a reliable and scalable integration between Creatio AI and the LLM, allowing for easy updates or replacements of components without affecting the entire system.
LLM
The LLM is one of the core components of Creatio AI. With the help of the LLM, Creatio AI can do the following:
- Understand user requests in natural language.
- Predict the current task of the user.
- Process input data.
- Generate relevant information for execution of actions.
- Invokes one ore more appropriate user actions.
The LLM ensures intelligent interaction with the user by leveraging advanced natural language processing algorithms.