Proficientai

Building sophisticated conversational AI applications typically requires significant development effort, from training custom models to managing backend infrastructure. However, Proficient AI streamlines this process through a powerful platform that lets developers create, manage, and integrate AI agents with minimal technical overhead. In this documentation, we'll explore how to set up and deploy Proficient AI agents, as well as the technical details of how these AI-powered conversational components work under the hood.

Overview and Setup

Getting started with Proficient AI involves three main steps: creating an agent on the web dashboard, customizing its configuration for your application, and connecting it to your app through a simple integration process.

To begin, navigate to the Proficient AI platform and create an agent using the specified model from the web dashboard. Once your agent is created, you can customize it according to your application's requirements by editing its configuration settings.

The integration process requires adding a prebuilt UI component to your app, which can typically be accomplished with 3-5 lines of code. This straightforward approach allows developers to quickly connect their applications to a conversational AI agent without significant backend development.

Proficient AI supports multiple types of applications, including business knowledge chatbots, virtual tutors, and AI assistants, enabling developers to create specialized conversational agents for various use cases.

Under the hood, Proficient AI operates cloud-based agents that communicate with the platform through a distributed architecture. While simple applications may not require a custom backend, agents can connect to various Large Language Models (LLMs) through the platform's infrastructure.

The platform's architecture consists of two primary API categories. The Client API provides access to a limited subset of operations, with examples including message sending capabilities. The Admin API, which requires admin-level access, offers complete agent lifecycle management functions such as creation, configuration, and deletion, along with tools for monitoring interactions and managing user data.

Each agent object includes essential properties like active status, creation timestamp, and unique identifier. The agent system maintains active and inactive states, where inactive agents prevent new message requests while preserving existing interactions. Developers can retrieve agent information using the Admin API, including details such as display name, description, and configuration timestamps.

Platform Architecture

The Proficient AI platform operates agents in the cloud, with each agent serving as a virtual entity managed through the platform's infrastructure. These cloud-based agents enable interaction with various parts of an application, connecting through the platform's distributed architecture.

The platform's operations are divided into two primary API categories, each with distinct capabilities:

Client API

The Client API provides basic functionality for client applications, allowing access to a limited subset of operations. While client applications can send messages to agents, they lack the capability to perform configuration tasks such as adjusting agent settings or managing the agent lifecycle.

Admin API

Backend applications interact with agents through the Admin API, which requires admin-level access for full agent lifecycle management. This API enables creation, configuration, and deletion of agents, offering comprehensive control over the agent's operations. Additionally, the Admin API provides tools for monitoring interactions and managing user data, allowing developers to maintain and optimize their conversational AI solutions.

Agent Properties and Lifecycle

Each agent maintains several key properties, including active status, creation timestamp, and unique identifier. The system manages agent states through active and inactive modes, where inactive agents prevent new message requests while preserving existing interactions. Developers can retrieve agent information using the Admin API, accessing details such as display name, description, and configuration timestamps.

The platform's architecture efficiently handles various application types, from business knowledge chatbots to virtual tutoring systems. While simple applications may not require a custom backend, agents can connect to multiple Large Language Models (LLMs) through the platform's infrastructure, enabling flexible integration options for different use cases.

API Fundamentals

The Proficient AI platform operates through two primary API categories, each serving distinct purposes in agent management and application integration.

Client API

The Client API enables basic operations for client applications, allowing them to interact with agents through a restricted set of functions. This API is designed to prevent client applications from performing sensitive operations like agent configuration, focusing instead on straightforward interactions such as sending messages.

Admin API

For comprehensive management of conversational AI agents, Proficient AI provides the Admin API, which requires admin-level access to fully utilize its capabilities. This API encompasses all aspects of agent management, including creation, configuration, and deletion. Additionally, the Admin API offers essential tools for monitoring agent interactions and managing user data, empowering developers to maintain and optimize their conversational AI solutions.

API Functionality

A detailed examination of the platform's API architecture reveals the specific operations supported by each category. The Client API grants access to a limited subset of functionality, with sending messages to agents representing the primary interaction point between client applications and the platform. In contrast, the Admin API provides unrestricted access to all agent lifecycle operations, enabling complete management of agent configurations and interactions.

Agent Properties and Management

The platform maintains several key properties for each agent, including active status, creation timestamp, and unique identifier. These properties enable developers to manage agent behavior through API calls, with the system maintaining active and inactive states to control interaction capabilities. When an agent is deactivated, it prevents new message requests while preserving existing interactions, providing a flexible mechanism for managing conversational AI operations.

The comprehensive API structure supports various application types, from specialized business chatbots to general virtual assistants. While simple applications may operate within the platform's infrastructure without requiring a custom backend, more complex implementations can leverage the Admin API's full capabilities to manage agent interactions and data management.

Agent Management

Each agent maintains several key properties, including active status, creation timestamp, and unique identifier. The system manages agent states through active and inactive modes, where inactive agents prevent new message requests while preserving existing interactions.

Agent Properties

• active: Boolean value indicating whether the agent is currently active. Inactive agents cannot handle new message requests.

• created_at: Timestamp of object creation, measured in milliseconds since the Unix epoch.

• description: Developer reference description for the agent.

• display_description: Public-facing description visible to users.

• display_name: Public-facing name visible to users.

• id: Unique identifier of the agent.

• name: Internal reference name designed for developers.

• object: Enum type indicating the agent object type (always "agent").

• updated_at: Timestamp of the last object update, measured in milliseconds since the Unix epoch.

Lifecycle Management

To manage agent operations, developers primarily use the Admin API, which requires admin-level access. Operations include:

• Creation of new agents

• Configuration changes

• Deactivation or deletion of existing agents

• Monitoring interactions

• Managing user data

The system enables developers to deactivate agents using the "Deactivate an agent" endpoint. This action prevents new message or interaction requests while keeping existing interactions intact. To perform this operation, the required parameter is the agent's unique identifier (agent_id).

Data Management

Agents store interactions and user data within the Proficient AI platform's infrastructure. Developers can access this data using the platform's APIs or through the admin dashboard. The platform manages all aspects of agent operations, dividing functionality into two primary API categories:

• Client API: Provides basic functionality for client applications, allowing message sending capabilities but not agent configuration.

• Admin API: Offers comprehensive management capabilities with unrestricted access to all agent lifecycle operations.