Toni: AI phone assistant with GPT real-time – developed for professional client conversations

Our assistant is self-developed – including AI prompts, routing, integrations, and dashboard. This allows us to fully control quality, latency, and the free choice of the best AI model in each case.

What Toni does for you in daily life

• Understand & clarify requests: Toni recognizes intentions such as *project inquiry*, *support*, *invoice* or *call back* and specifically asks for missing information if necessary.  

• Provide assistance directly or connect: Answer recurring questions, handover to the right contact person (warm/cold transfer), or organize appointments and callbacks.  

• Summarize & document: Structured notes, contact details, and context land directly in your systems – traceable and neatly filed.  

• Scalable & reliable: Whether peak times or outside business hours: Toni remains consistent, polite, and efficient.  

• DACH-focused: Tone and word choice fit – German first, additional languages optional.

This allows you to qualify inquiries faster, reduce processing effort, and improve customer experience. Call and test Toni live: +43 677 61279177

Architecture in a Nutshell: Realtime AI Meets Telephony

Under the hood, Toni connects Twilio with OpenAI gpt-realtime. This results in low latencies and you can interrupt Toni without losing the flow of conversation. Our orchestration layer controls call flows, policies, and forwarding; the Dashboard provides real-time transparency.

Technical Components (Excerpt):

• gpt-realtime: Streaming understanding, naturally sounding responses, interruptibility

• Twilio Voice: Call acceptance, DTMF/IVR bypass, transfers, number management

• MCP Layer: Controlled tool accesses (read/write) with policies & versioning

• Dashboard: Live transcript, routing, KPI monitoring, roles & permissions

Dashboard: Control and Quality in Real-Time

Our Dashboard bundles the relevant signals of an ongoing conversation – in a clear, auditable form:

• Live transcript with key phrases and automatically created summary

• Routing rules (opening hours, priorities, employee lists)

• Roles & permissions for clear responsibilities

MCP Integrated: Connect, Reconcile, Document

Toni uses the MCP (Model Context Protocol) to form structured connections with various systems. This enables:

• Slack notifications & handover: Transfer along with conversation summary to the appropriate channel.

• CRM updates: Create or enrich leads/contacts; set status.

• Ticketing: Create support tickets with priority, category, and notes.

Data Protection & Hosting (EU/GDPR)

We rely on EU hosting and data minimization. Accesses are role-based. Sensitive content can be masked in the transcript. The goal is a revision-proof, transparent operation.

Who Toni is Well-Suited For

• Sales & Business Development: Pre-qualification, needs analysis, appointment scheduling – without losing leads.

• Customer Service: First-time solutions for standard requests, structured escalation, better accessibility.

• SMEs & Scale-ups in DACH: Professional call handling without 24/7 staffing, clearly measurable benefits.

A selection of our work can be found in the projects.

Why In-House Development Makes the Difference Here

Because we built Toni entirely ourselves, we can:

• choose the best model – balance quality versus cost, switch models, iterate quickly;

• quickly address edge cases and swiftly map new call patterns;

• build knowledge and provide individual support in complex cases.

Try It Out Immediately

Call and test Toni live: +43 677 61279177
If you want to use Toni for your company, call him – we'll clarify the next steps directly on the phone.