A New Way to Get Answers: Multiple Voices, One Conversation
We’re excited to introduce the Dynamic Circle of Speakers — a first-of-its-kind multi-speaker AI pipeline that brings together a curated group of distinct voices to respond to your questions, prompts, and conversations.
Instead of hearing from just one AI model, the Circle assembles a living panel of speakers — each with their own personality, expertise, and perspective — and orchestrates them in real time to give you richer, more thoughtful answers.
How It Works
Intelligent Speaker Orchestration
When you send a message, the Circle doesn’t just pick speakers at random. An AI-powered orchestration engine analyzes your message, considers the conversation history, and determines which speakers are most relevant — and in what order they should respond.
- Context-aware ordering — The orchestrator reviews recent conversation history (configurable depth) to understand the flow and select the best voices for each round.
- Tag-based filtering — Speakers are organized by tags (e.g., theology, counseling, technology). Use simple tag queries to focus the conversation: combine tags with commas for OR logic, or
+for AND logic. - Dynamic selection — If no tags are specified, the orchestrator intelligently selects from all available speakers based on the topic at hand.
Real-Time Conversation Awareness
Every speaker in the Circle is aware of what the others have said. The second speaker sees the first speaker’s response. The third sees both. This creates a natural, flowing discussion where each voice builds on the last — not a disconnected series of isolated replies.
- First Speaker Prompt — The opening speaker receives a special directive to set the tone and address your question directly.
- Subsequent Speakers — Each following speaker receives a summary of prior responses so they can add unique value, offer a different angle, or respectfully build on what’s been said.
Dynamic Introductions
The first time a speaker participates in a conversation, they’ll introduce themselves — sharing their name and area of expertise in a concise, natural way. After that, they jump straight into the discussion. This gives you a sense of who’s speaking and what perspective they bring, without repeated introductions.
Personalized Responses
The Circle integrates with your user profile to personalize responses. If you’ve shared information about yourself in your system prompt — your background, interests, or preferences — speakers will use that context to make their responses more relevant and personal to you.
Profile Images & Visual Identity
Each speaker has a profile image displayed alongside their response, so you can visually identify who’s speaking at a glance. The Circle supports configurable image sizes and custom image hosting.
Streaming with Live Status
Responses stream in real time with live status indicators showing which speaker is currently thinking. You’ll see:
- The number of speakers selected and their names
- The AI-determined speaking order
- Real-time “thinking” indicators as each speaker formulates their response
- A “Circle Complete” confirmation when all speakers have responded
Tool Integration
Speakers can be configured with tools — external capabilities like web search, Bible lookup, or other integrations. When a speaker uses a tool, you’ll see inline feedback showing which tools are being employed, along with proper citations.
Thinking Transparency
For models that support reasoning (like chain-of-thought or thinking models), the Circle renders their internal reasoning in collapsible detail blocks — so you can optionally peek behind the curtain to see how a speaker arrived at their answer.
Built on a Solid Foundation
The Dynamic Circle v2.0 is built on the OpenWebUI Streaming Blueprint — a reusable, battle-tested foundation that handles API communication, model resolution, streaming, and tool integration. This means the Circle benefits from:
- Reliable streaming with proper error handling
- Automatic model ID resolution (speaker names map to their underlying models)
- Token usage tracking across all speakers in a single conversation
- Memory-aware context injection (web search results, memory context, and more are shared with all speakers)
What Makes the Circle Special
| Feature | Traditional Chat | Circle of Speakers |
|---|---|---|
| Perspectives | Single model | Multiple curated voices |
| Awareness | Isolated responses | Each speaker hears the others |
| Ordering | Fixed | AI-orchestrated per question |
| Personalization | Basic | Profile-aware, context-rich |
| Introductions | None | Dynamic, first-time only |
| Tools | Per-model | Per-speaker with citations |
| Streaming | Standard | Multi-speaker with status |
Who Is It For?
The Circle of Speakers is designed for anyone who values depth, diversity of thought, and meaningful conversation — whether you’re exploring Scripture, seeking counsel on a life decision, learning something new, or simply looking for a richer AI experience.
It’s not about more words. It’s about more wisdom.
Getting Started
The Dynamic Circle is available now on the Cross & Faith AI platform. Simply select the Dynamic Circle pipeline from your model list and start a conversation. The speakers will introduce themselves and the discussion will begin.
Your administrator configures which speakers participate by default. Depending on the topic or group, they may set up multiple Circle pipelines — each tuned to a different set of voices using tag-based filtering. For example, one Circle might focus on theology and apologetics, while another brings together counseling and encouragement voices. Just pick the Circle that fits your conversation.
Built with care by Beau D’Amore for the Cross & Faith community.