Turning Chatbots into Phonebots (without breaking everything)
Product & UX
You’ve built a chatbot. Great. Then someone asks: “Can this work on the phone too?”
In this workshop, we’ll show how to turn chatbots into phonebots — and what really needs to change to make that work well. We’ll explore how one system can power multiple channels, why that’s practical, and where the limits are.
Technically, moving from chat to voice is often simpler than expected. The real challenge is elsewhere: voice isn’t just chat with sound. It comes with different constraints, failure modes, and user expectations.
That’s why we cover two perspectives:
Engineering: Connecting bots to real phone systems via APIs, handling calls and events, and understanding which chatbot components translate to telephony — and which don’t.
Conversation & UX Design: Designing for interruptions, hesitations, and mid-sentence changes without creating awkward experiences.
Compact, practical, and based on real-world telephony systems.
Time & Place
Wed, March 25
14:00 - 15:30
The Ritz-Carlton Berlin
Salon Bellevue
Classroom Seating
Max. Capacity: 45 Seats
Meet Your Intructors
Laura Grimm
Product Manager, sipgate
Laura is a Product Manager in the AI Innovations team at sipgate, where she works on technically complex AI systems for communication products. The team explores new approaches beyond language models, combining machine learning, data processing, and real-time systems to develop and test new capabilities.
In her role, Laura focuses on helping shape these technologies into reliable and scalable product solutions. With a background in product management, conversational UX, and user research, she contributes to connecting technical innovation with clear product direction and practical use cases.
Stefan Lange-Hegermann
Product Lead, sipgate
AI Evangelist & Product Lead AI, sipgate GmbH · Stefan Lange-Hegermann explores the question of how product development changes when code costs almost nothing anymore. Prototypes replace specifications, building replaces discussing — and judgment becomes the real bottleneck. He also works on this in practice at sipgate: as Product Lead for sipgate Flow, a platform that enables AI agents to conduct real phone calls — with phone numbers, call control, audio streams, STT and TTS via an API, without needing your own telephony infrastructure.
What To Expect
Who Is This For?
Product Managers
Conversation Designers
Voice Engineers
Solution Architects
CX Leaders
Pre-Requisites
No specific technical skills required
Bring Your Own Laptop
What You'll Learn & Do?
Chat vs. voice: key differences
What translates — what breaks
Connecting bots to telephony APIs
Designing for interruptions and latency
Avoiding awkward voice experiences
Agenda & Activities
Agenda for this session:
Getting Settled (5 minuten)
Build a Chatbot (hands-on) Create a simple text-based chatbot. (10 minutes)
Designing for Voice (theory) What changes when text becomes speech — pitfalls & how to solve them. (15 minutes)
Reprompt for Voice (hands-on) Take your chatbot and make it work for voice. (20 minutes)
Break (5 minutes)
Getting to the Phone (theory) Latency, VAD, events, routing & call transfers. (15 minutes)
Build Your Own Voice Bot (hands-on) Put it all together. (20 minutes)
Reflection / Discussion (10 minutes)
Prerequisits:
Basic understanding of HTTP and programming fundamentals
First hands-on experience with “vibe coding”
Bring Your Own Laptop
.png)