you read data
Call records, live calls, extensions and users. Authenticated with your organization code plus an IP allow-list in CPanel.
call-log { from, to }What is an AI Tool (Tool Calling)?
A language model on its own can only produce words. A tool gives it hands: a defined function — a name, the inputs it expects, and a result — that the agent calls mid-conversation to fetch live information or take a real action. Voicenter ships fourteen of them.
tool click2call { in: { phone, target } → out: { CALLID } }How Voicenter tools behave
The same fourteen tools, grouped by how they behave and how they authenticate.
you trigger something
Place a call, set agent status, edit the blacklist, drive a dialer campaign. Same code plus IP auth.
click2call { phone, target }Voicenter calls you
Voicenter POSTs to an endpoint you host, configured in CPanel or in the Voice Agent intent. No outbound token — respond fast.
popup-screen → your endpointa continuous feed
Live call, queue and agent-status events streamed over socket.io, or a JS tag that attributes calls to campaigns.
real-time (socket.io)the same surface, no tokens
The Voicenter MCP server exposes all of it to any MCP client. You sign in over OAuth — nothing is stored in a .env file.
mcp01.voicenter.co/mcpdesign, not runtime
The Bot Builder pipeline designs a voice agent once and emits deployable JSON. It never sits in the live call path.
bot-<name>-<date>.jsonKnowledge, tools, and MCP
Three ideas that are easy to mix up — kept clear.
Knowledge = information
What the agent knows in advance — products, procedures, FAQs. It helps the agent answer correctly, but not act.
docs · FAQs · proceduresTool = action
The agent can instantly pull call records, place a call, update the CRM, or push a screen-pop — while the conversation is still running.
call-log · click2call · cdr-notificationMCP = one contract
Expose a tool once over the open Model Context Protocol and any supporting agent can use it. Voicenter runs one at mcp01.voicenter.co/mcp.
define once and use anywhereHow a tool call works — in five steps
From the customer's words to a done action — the loop every agent runs, in seconds.
Intent detected
“Call this lead back now”Tool selected
click2call { target }System executes
POST /ForwardDialer/click2call.aspxResult returned
ERRORCODE: 0 · CALLIDAgent responds
“Ringing your rep now 📞”CATEGORY EXPLAINER
What is a conversational AI voice agent — and how do you build one?
A conversational AI voice agent (a voicebot) is software that answers or places phone calls and holds a real spoken conversation: it understands what the caller wants, collects the information it needs, takes actions in your systems, and routes the call. In an AI call center it handles routine calls end to end — booking, lookups, status, qualification — and hands the rest to a human, so wait times drop and agents focus on the calls that need them.
How to build a voice agent — three steps
Define the structure
Decide the agent's identity and persona, the intents it handles, and the data (slots) it needs to collect for each one. This is the skeleton of the bot.
voicenter-bot-spec-designerWrite the language
For every intent, write what the agent says to collect and confirm each piece of information, and what it does once it has it — in natural, on-brand speech.
voicenter-bot-intent-detail-authorValidate & deploy
Turn the completed design into the platform's wire format, validate that everything is consistent, and ship the resulting JSON. The agent is then live.
voicenter-bot-json-assemblerContinue from here
See tools in the hands of real agents, or browse the full catalog.
Common Questions
The questions people ask about tool calling
KEEP EXPLORING
See tools in the hands of real agents
Three scenarios showing agents orchestrating tools end to end.
14 tools · Calls · Records · Agents · Campaigns · Real-time · Inbound