Concept
Dashboard tour
Know where everything lives in the dashboard: agents, numbers, tests, conversations, keys and settings.
Everything you build meets in one dashboard at https://vlayers.ai/dashboard. You build an agent, test it in the Playground, put a number on it, and read every call it takes. The Agents list is the hub: however an agent was made (canvas, code or connector), it lands there.
Build
Where agents and their parts live.
| Page | What it is for |
|---|---|
| Agents | Every agent in one list: flow deploys, playbook deploys, SDK agents (with live worker status) and brain connectors. Each agent has its model, voice and pipeline settings. Its Versions tab holds the draft and live configurations: promote a draft to live, split new calls between them, and see published versions. |
| Flows | The visual builder. See Flow builder. |
| Playbooks | Reusable, goal-driven skills. Version them, publish them, and deploy them alone or use them from flows. See Playbooks. |
| Tools | What your agents can call: HTTP tools you define, and the tools of MCP servers you connect, one tab each. See Tools. |
| Knowledge base | The knowledge bases behind the Knowledge step. |
Deploy
Where people reach your agents.
| Page | What it is for |
|---|---|
| Phone numbers | Search, buy and attach numbers to agents. Inbound calls go straight to the assigned agent. See Phone numbers. |
| Channels | WhatsApp, Telegram, Discord and SMS registration, and which agent answers each number, bot or server. |
| Outbound calls | Run a scheduled batch campaign from a CSV, place a single call, or start calls from call triggers. Every call passes the do-not-call and consent rules first. |
Test
Prove it works before it takes a call.
| Page | What it is for |
|---|---|
| Playground | Try any agent from the browser by voice, text or chat. No number, no phone setup. |
| Tests and evals | Test suites run scripted conversations with pass/fail checks per turn. Call scoring has an AI judge rate real calls, charged to your wallet. |
See Testing.
Monitor
What happened on every call and conversation.
| Page | What it is for |
|---|---|
| Conversations | Every conversation on every channel, a tab per channel. Under Voice, each call opens to its transcript, recording, cost and latency. |
| Analytics | Call volume, outcomes, cost and latency over time. |
Developers
What you use when you build on VoiceLayer from your own code.
| Page | What it is for |
|---|---|
| API keys | Create keys for your servers and SDKs, MCP clients, the chat widget and agent workers. See Authentication and API keys. |
| MCP server | Give AI clients (Claude Code, Cursor, Codex, claude.ai, ChatGPT) your agents, with copy-paste setup and a scoped key. See MCP. |
| Custom LLM | Put a voice on your own LLM. See Brain connectors. |
| Webhooks | Endpoints VoiceLayer calls when something happens. See Receive with webhooks. |
| Data export | Forward LLM and telemetry data to your own observability tools. |
Settings
What you set once.
| Page | What it is for |
|---|---|
| Calling | Caller ID, outbound trunk and recording settings for the workspace. |
| Variables | Workspace variables shared by every flow. See Variables. |
| Provider accounts | The AI model, texting and calling accounts your agents use, whether each one works, and what depends on it. Secrets are encrypted and never leave the server. |
| Transfer destinations | The numbers code may transfer calls to. See Transfers. |
| Team and roles | Invite members, set their roles and review access requests. |
| Billing | Plan, usage and invoices. |