Multi-agent personas
Emma at reception, Julia in sales, Thomas in support: multiple agents with their own voice, role and instructions — running in parallel.
AI telephony · GDPR-compliant · Self-hosted
Voice Agent Studio is the multi-tenant platform for AI phone agents: define, configure, protect, monitor, deploy — without writing a line of code, on your own infrastructure.
01 //The problem
Every missed call costs revenue. Traditional phone menus frustrate callers. And generic AI bots are a black box no one wants to defend in front of their data protection officer. Voice Agent Studio solves all three at once:
Hold queues and missed calls
AI agents answer every call instantly — around the clock, in natural language.
AI as an uncontrollable black box
Every agent operates under explicit rules and guardrails; every change is versioned and traceable.
Data protection as a dealbreaker
Consent management, deployment on your own infrastructure, and full data sovereignty — GDPR is built in, not bolted on.
02 //Features
Every feature listed here is in production today — no vaporware promises.
Emma at reception, Julia in sales, Thomas in support: multiple agents with their own voice, role and instructions — running in parallel.
Inbound and outbound trunks with dispatch rules. Connects to your existing phone system — no number migration required.
Agent-to-agent and agent-to-human handovers, warm or cold — with keyword and rule matching to reach the right person.
Agents take real actions: book appointments, query your CRM, trigger n8n workflows. Editor and test interface included.
Connect your company knowledge: the agent answers from your documents and data — not from the model’s gut feeling.
OpenAI Realtime, Azure OpenAI, Gemini Live, or your own pipeline with Deepgram, ElevenLabs, Anthropic, Ollama — switch freely.
Recording and transcription only with the caller’s documented consent — requested and logged automatically.
PIN, caller-ID matching or CRM verification before the agent touches sensitive processes.
Strictly isolated configurations per tenant — for corporate structures, agencies and managed-service scenarios.
Active calls in real time, KPIs for calls, duration, tokens and cost — plus a full history with transcripts.
Every prompt and configuration change is versioned, diffable and can be rolled back at any time.
Role-based access control for teams — and an MCP interface with 33 tools to administer the studio straight from LLM clients.
03 //Governance
The core architecture of Voice Agent Studio: a strict rule hierarchy with downward inheritance and upward immutability. The outer layer always wins.
Apply to all agents. Immutable at agent level. Sourced from company policy, compliance and GDPR.
Technical safeguards with moderation and compliance settings. They intervene automatically.
Configurable per agent. They extend the company rules — but may never contradict them.
“An agent can change its work instructions. But it can neither break the constitution nor switch off the police.”
04 //Trust
Consent management, data exports and deletion concepts are product features — not afterthought add-ons.
Developed and supported by ridon IT GmbH in Baden-Württemberg. Contracts, support and data processing under German law.
Runs via Docker on your server or in your private cloud — data stays where you decide.
Multi-provider architecture and full configuration exports: you can switch at any time — even away from us.
05 //Who it’s for
06 //Contact
30 minutes, your use case, a real phone call: we’ll show you how a voice agent works under your rules.
Or write directly toinfo@ridon.de