I am trying to create a chat agent, how will i go about it?

Thanks for reaching out, Aditya. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

is the python based and no-code agent builder offer same features and controll overt the bot?

Good question — here's a concise comparison so you can pick the right path. What you get with the Python SDK (code-based) • Full programmatic control: implement custom Agent classes, override lifecycle hooks, and add complex logic. • Complete access to STT/LLM/TTS/VAD selection and configuration per session. • Custom plugins, prewarm routines (e.g., load models on startup), and low‑level session handling. • Easier to integrate arbitrary external services, custom data stores, or advanced telemetry. • Best when you need fine‑grained control, custom behavior, or production integrations. What you get with the Agent Builder (no-code) • UI-driven setup: set instructions, welcome messages, choose models, and wire simple HTTP tools. • Fast prototyping and testing without writing code. • Good for standard chat/voice agents and quick iterations. • Limited or no access to deeper customization (custom plugins, fine-grained session hooks, complex prewarm logic). • Best when you want speed and simplicity over full customization. Recommendation • Choose Python SDK if you need full flexibility, custom workflows, or advanced integrations. • Choose Agent Builder if you want to prototype quickly or manage simpler agents via a UI. Would you like a short Python starter example or a walkthrough of the Agent Builder UI?

yes how can i do ti with Agent Builder (no-code)

Great — I’ll walk you through creating a chat agent with the Agent Builder (no-code). I’ll cover accessing the Builder, configuring the agent, testing, and deploying. 1) Open the Agent Builder • Sign in to your LiveKit dashboard and open the Agent Builder (browser-based tool). 2) Create a new agent • Click “Create agent” (or “New agent”). • Give it a name and optional description. 3) Set core behavior • Instructions: Enter the agent’s system prompt or high-level behavior (how it should respond, tone, limits). • Welcome greeting: Add the message users see when a session starts. 4) Choose models and capabilities • Model selection: Pick the LLM, TTS, and STT models you want the agent to use (or leave defaults). • Enable voice or chat mode depending on whether you want text-only or voice-capable agent. 5) Add tools / integrations • HTTP Tools: Configure simple HTTP endpoints the agent can call (set method, URL, headers, and sample payload). Use these to fetch data or call external APIs without code. • Variables / context: Configure any persistent context fields or slots the agent should track (if available in the UI). 6) Control response behavior • Response settings: Adjust temperature, max tokens, or other model parameters exposed in the Builder. • Safety / constraints: Add guardrails in the instructions or using the Builder’s settings to limit output or enforce policies. 7) Test the agent • Use the in‑Builder test console to send messages or join a test room for voice agents. • Iterate: update instructions, tools, or model choices based on test results. 8) Deploy and connect • Save and publish the agent. • Deploy options: create a deployment or get the agent’s integration endpoint/ID to connect from your app or a room. • For voice agents, follow the Builder’s instructions to add the agent to a LiveKit room. 9) Monitor & update • Use Builder logs or dashboard testing results to review conversations. • Update instructions, tools, or model settings and republish when you need changes. If you want, I can: • Provide a sample set of instructions/welcome message for a helpful customer‑support agent. • Show exactly which model and HTTP tool fields to set (I can list typical values). • Walk through the UI step‑by‑step with screenshots if you tell me which part you’re on now. Which of these would help most?

but the pipeline only shows STT-LLM-TTS or Realitime models, how to create a chat agent?

You're right to question that — the pipeline visualization shows STT→LLM→TTS, but you can create a chat-only agent in the Agent Builder by using only the LLM/text portion. Here’s how to do it in the Builder (no code): 1. Open Agent Builder and create a new agent. 2. In Core settings: ◦ Set your agent Instructions and Welcome greeting. ◦ Choose the LLM model you want. 1. Disable audio: turn off voice/STT and TTS options (or select “text only” / uncheck audio features) so the agent uses text input/output only. 2. Add tools/integrations if needed (HTTP tools, variables). 3. Configure response controls (temperature, max tokens) and safety guardrails. 4. Test in the Builder console — send text messages and confirm replies appear in the text test window. 5. Publish/deploy and connect your app to the agent’s text endpoint or add the agent to a room with audio disabled. Notes: • The Builder handles the text-only flow behind the scenes; you don’t need STT/TTS for chat agents. • If your frontend needs to receive agent text responses, listen for the transcription/text stream (the Builder’s test UI shows this automatically). Would you like a suggested starter Welcome message and instructions for a customer-support chat agent? Sources: Text and transcriptions | LiveKit docs | How to Set Up Speech-to-Text Input with Text-Only Output in LiveKit Agents | https://blog.livekit.io/announcing-agent-builder-a-pathway-to-code/#/portal|

how to turn off voice/STT and TTS options

Thanks — I’ll show how to disable voice/STT and TTS in the Agent Builder so your agent is text‑only. 1. Open the Agent Builder and select your agent (or create a new one). 2. Go to the agent’s Settings or Capabilities panel (often labeled “Channels”, “Media”, or “Capabilities”). 3. Turn off or uncheck any options labeled Voice, STT, Speech-to-Text, Microphone, or Enable audio input. 4. Turn off or uncheck any options labeled TTS, Text-to-Speech, Voice Output, or Enable audio output. 5. Save or publish the agent. If there’s a specific toggle for “Text-only” mode, enable that instead of toggling individually. A couple implementation notes: • With audio disabled, the Builder will use the LLM/text path only; responses are delivered as text. • If you plan to connect this agent to a LiveKit room, ensure room options disable audio (no audio publish/subscription) so the agent communicates via text streams. Tell me which screen or option you see in your Builder UI if you want step-by-step clicks or I can provide a suggested welcome/instructions for a chat-only agent.

I was reading about WebRTC transport and data i found streaming section, guide me how can i create a agent to test it fast on my local, given i have my hosted livekit server.

Thanks — I’ll help you spin up a quick local test that uses WebRTC/data streaming with your hosted LiveKit server. A couple quick clarifying questions so I tailor the steps: 1. Do you prefer Python or Node.js for the agent code? 2. Do you already have a published Agent (Agent Builder) or do you want a minimal custom agent using the SDK? If you want Python, I’ll give a compact runnable example that: • Installs the LiveKit agents package, • Shows an AgentServer + AgentSession that connects to your LiveKit server, • Listens for and sends data messages over the room data channel (WebRTC/data), • Explains how to run locally and test from a browser (LiveKit test client or minimal HTML page). Which language and agent type do you want me to proceed with?

I want python code

I don't have the answer you're looking for. You could also try asking your question: • in one of the other Slack channels or • to https://deepwiki.com/livekit/livekit_composite which is trained on all LiveKit source code If you find the answer, please post it here to help others!

generate python code to test on my local