From Bright Pattern Documentation

Revision as of 01:48, 7 November 2026 by Laura.donovan (talk | contribs) (Created page with "= AI Agent 2.0 Release Notes = '''''Bright Pattern AI Agent 2.0''''' is a powerful and expansive reimagination of the AI Agent scenario block, featuring LLM-driven orchestration, essential tools like retrieval-augmented generation and MCP connectors, and native integration with BPCC voice and chat scenarios. == Access == '''Requirements:''' * BPCC 5.36.0 or later The AI Agent 2.0 application is accesse...")

< Previous | Next >
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

AI Agent 2.0 Release Notes

Bright Pattern AI Agent 2.0 is a powerful and expansive reimagination of the AI Agent scenario block, featuring LLM-driven orchestration, essential tools like retrieval-augmented generation and MCP connectors, and native integration with BPCC voice and chat scenarios.

Access

Requirements:

  • BPCC 5.36.0 or later

The AI Agent 2.0 application is accessed independently from the Bright Pattern Contact Center application. To enable AI Agent 2.0 for your contact center, contact Bright Pattern Customer Success.

AI Flow Builder

The AI Flow Builder serves as a visual workspace for designing AI-native workflows.

Rather than hard-coding all decision paths and execution logic, you describe objectives, provide AI Agents with instructions, knowledge, and tools, and establish guardrails. During runtime, the AI model of your choice dynamically determines how to accomplish the task, orchestrating the right sequence of actions based on the business context and enforced constraints.

AI Flows consist of one or many AI Agents, each with their own scope of responsibilities and role in the overall automation process.

Within an AI Flow, AI Agents can be chained linearly, performing sequential steps in a process. For example, the first AI Agent in a sequence may handle mandatory authentication of the customer; then, based on the result of that agent's work, the AI Flow either passes the conversation to the next agent in the chain (authentication succeeded) or escalates the conversation to a human agent (authentication failed).

Additionally, AI Flows can be used to automate more complex processes via multi-agent orchestration. Multi-agent orchestration allows one AI agent to invoke other specialized AI agents, much as a human agent might transfer a customer interaction to a different team.  

Tool Suite

Custom API tools, integrations, and Model Context Protocol (MCP) servers grant AI Agents access to enterprise applications, external tools, and knowledge sources, allowing them to retrieve information, update records, execute business processes, and complete tasks.

Knowledge Bases, static documents, cloud repositories, and website content can be connected to AI Agent 2.0, allowing you to index, publish, and manage approved sources from one workspace. Retrieval-augmented generation (RAG) allows AI to retrieve accurate, company-aligned information at runtime rather than relying solely on what the underlying model already knows.

Integrated Testing

The AI Flow Builder contains a debugging interface where users can experiment, validate behavior, and refine AI workflows during development. Users can interact with the AI Agent in real time via mock tool inputs and a built-in chat interface to understand how it reasons, which tools it invokes, and what information it retrieves to support its decisions.

Evaluation

AI Evaluators are specialized AI components that continuously assess responses and actions before they are executed. Rather than relying on static validation rules, Evaluators apply business policies, quality standards, and enterprise-specific criteria to determine whether the AI's proposed action is acceptable.

If an issue is identified, the Evaluator provides targeted feedback, allowing the AI to refine its reasoning and try again. This evaluation loop continues until the response meets the required standard or the configured retry limit is reached, at which point the AI Flow can take an alternative action, such as escalation.

Bright Pattern Integration

Added AI Flow Scenario Block

The new AI Flow scenario block allows voice and chat scenarios to call complex AI Flows defined in the AI Agent 2.0 interface.

Added Scenario Variables

The following variables have been made available for scenarios featuring the AI Flow block:

  • $(aiFlowFinalMessage) - final message received from AI Agent 2.0 upon AI Flow completion
  • $(aiFlowErrorCode) - error code received from AI Agent 2.0 in the event of a Failed exit
  • $(aiFlowErrorMessage) - technical error message received from AI Agent 2.0 in the event of a Failed exit
    < Previous | Next >