A voice agent that resolves common tickets
Attach a knowledge base, collect the issue details, call internal APIs, and send a webhook when the session ends.
Knowledge fetch
Userdata capture
Webhook delivery
Talqing no-code agent builder
Voice and video agentswithout code or infra.
Design the agent, connect tools and knowledge, test live sessions, and publish a versioned realtime agent. Talqing handles the LiveKit runtime, session records, webhooks, usage, and operational glue.
Built for teams replacing manual conversations
Customer support
Sales qualification
Scheduling
Onboarding
Internal ops
Product coaching
Customer support
Sales qualification
Scheduling
Onboarding
Internal ops
Product coaching
From agent idea to live session, without a worker repo.
Talqing turns no-code configuration into a running realtime agent, then keeps every call observable and versioned.
Builder inputNo code
Brief
Describe the job
Definition
Prompt, models, channel
Talqing compile layer
Everything a LiveKit agent needs, modeled visually.
Models
Tools
Knowledge
Hooks
Secrets
Webhooks
Live outputVersioned
Voice
LiveVideo
LiveText
LiveThe workflowno code path
One path from first prompt to published realtime agent.
Talqing keeps agent authoring, runtime configuration, testing, and publishing in one place, so teams can improve behavior without shipping a new service.
publish review
Agent ready to publish
Immutable version will be created
Ttalqing/prompt- instructions and behavior readyreq12s
Ttalqing/channel- voice + video settings validreq08s
Ttalqing/tools- 7 published tool definitionsreq19s
Ttalqing/knowledge- 2 knowledge bases attachedreq15s
Ttalqing/runtime- version snapshot frozenreq22s
published version v12Live
runtime dispatch ready
Capabilities
The pieces users expect from a serious agent builder.
Talqing is built around the same primitives engineers use in LiveKit agent code, but exposed as product controls that non-engineers can actually operate.
Voice agents
Realtime browser calls with STT, LLM, TTS, VAD, interruption settings, transcripts, and usage tracking.
Video agents
Avatar-backed sessions for face-to-face experiences, configured from the same no-code agent model.
Copilot builder
An embedded AI builder that can update prompts, models, tools, hooks, knowledge, and publish state.
No-code tools
Compose HTTP calls, code steps, frontend RPC, MCP/provider operations, handoffs, and webhook actions.
Knowledge bases
Crawl, review, build, and attach knowledge so agents can fetch grounded answers at runtime.
Session observability
Review calls, transcripts, tool invocations, final userdata, provider usage, and cost breakdowns.
3
Voice, video, and text channels
Frozen
Published agent versions
LiveKit
Compiled runtime sessions
Usage
Per-session cost breakdowns
Who it is for
Built for teams that own conversations, not infrastructure.
Operations teams
Turn repeatable customer conversations into agents that can answer, collect data, trigger actions, and hand off when needed.
Product teams
Prototype a new conversational experience without waiting for a custom LiveKit worker or tool integration sprint.
Support leaders
Give customers a voice-first front door backed by your docs, CRM actions, and clear call history.
Engineering teams
Keep control over models, tools, hooks, webhooks, secrets, and runtime behavior without becoming the bottleneck for every edit.
Agent patterns
Real workflows, not another chat wrapper.
Voice and video agents become valuable when they can remember session state, take actions, and leave an auditable trail.
A qualification agent that feels like a real rep
Ask adaptive questions, store lead fields, trigger a CRM action, and hand off to another agent when the buyer is ready.
No-code tools
Agent handoff
Call transcript
An avatar guide for onboarding or demos
Use the same agent definition with a video channel, a selected avatar, and frontend actions for interactive UI moments.
Avatar channel
Frontend RPC
Live testing
FAQ
Questions, answered.
What can I build with Talqing?
Talqing is for building voice, video, and text AI agents that can talk to users, use tools, retrieve knowledge, call webhooks, update session state, and hand off to other agents. The core focus is realtime voice and video agents.
Do I need to write LiveKit Agents code?
No. You author an agent definition in the dashboard or through Talqing Copilot. Talqing compiles that definition into a LiveKit AgentSession and runs it through the platform runtime.
What does Talqing handle for me?
Talqing handles the agent runtime, published versions, session dispatch, transcripts, tool execution, webhooks, knowledge retrieval, usage collection, and cost breakdowns. You focus on the agent behavior and connected business systems.
Can I test an agent before publishing?
Yes. You can run browser voice, text, and video test sessions, then inspect the transcript, tool calls, userdata, usage, and cost for each session before publishing a version.
How do tools work without code?
Tools are built as operation trees. You can configure HTTP requests, conditional branches, code operations, generated replies, frontend RPC, MCP/provider operations, handoffs, end-call actions, and webhook-triggered flows from the UI.
Can the agent use my product knowledge?
Yes. Talqing can crawl and build knowledge bases, let you review the extracted pages and structure, and attach those knowledge bases to agents so runtime calls can fetch grounded context.
Is telephony included today?
Yes for bring-your-own Plivo and Exotel. Connect a carrier account under Phone Numbers, import DIDs, assign a published voice agent for inbound, and place outbound test calls from the dashboard. PSTN billing stays with your carrier; Talqing runs the agent. Twilio/Telnyx, transfers, DTMF, and batch dialers are planned next.