AI Assistants

Overview

AI Assistants introduce intelligent automation capabilities directly within the Flow Builder, enabling users to create, modify, and troubleshoot flows using natural language.

This release includes two AI-powered assistants:

Flow Generator –

Creates new flows or modifies existing flows based on user instructions.

Debug Inspector –

Analyzes flow executions to identify issues, investigate errors, and provide insights using logs, metrics, and execution traces.

Although these capabilities are experienced as AI-powered features, they are implemented using a new concept called System Flows.

---

Understanding System Flows

What Are System Flows?

System Flows are specialized flows that power AI Assistants behind the scenes.

They use the same building blocks available in standard flows, including:

• Nodes

• Agents

• Goals

• Branching logic

• Runtime execution engine

Each AI Assistant is backed by a dedicated System Flow that executes its behavior.

Key Benefits

• Built using the platform's own flow technology

• Runs on the same runtime engine as standard flows

• Reusable and maintainable through version control

• Automatically available across all organizations

---

Deployment Process Step-by-Step

YFlow YAML Files in Source Control

Developers create and maintain System Flows as YFlow YAML files.

• The flow definitions are stored in the platform's Git repository.

• These files contain the complete configuration and logic of the AI assistant.

• Changes are version-controlled and managed like any other software code.

Example: A Flow Generator System Flow is defined as a YAML file and committed to the repository.

Infrastructure Deployment

When a new platform release is deployed:

• The infrastructure deployment process packages the System Flows.

• The flow definitions become part of the deployment pipeline.

• The latest versions are prepared for distribution to the target environment.

Uploaded to S3

During deployment, the System Flow files are uploaded to Amazon S3.

• S3 acts as the central storage location for System Flows.

• Each environment can access the same centralized flow definitions.

Benefit: A single source of truth exists for all System Flows.

Configuration Service Loads Flows at Startup

When platform services start:

• The Configuration Service retrieves the System Flows from S3.

• The flow definitions are loaded into the platform configuration.

• The System Flows become available within the runtime environment.

This means no manual installation is required by individual organizations.

Available to Every Organization

Once loaded:

• All organizations in that environment automatically receive access to the same System Flows.

• The AI assistants become available without additional setup.

• Organizations do not need to import, create, or configure these flows themselves.