Thesis: Design and evaluation of a voice AI agent for 24/7 structured support intake
Knightec Group Sweden
- Sundsvall, Sweden
- Mid Level
- Bachelor's degree
- Posted
Skills
Required
- telephony
- STT
- LLM
- TTS platforms/APIs
Preferred
- scripted
- LLM-driven
- flow
High Level Description Customer Success and Support teams are expected to be reachable around the clock, but staffing phone lines 24/7 is expensive and often unrealistic. Outside business hours, customers are typically met by voicemail or a web form, which leads to incomplete information, delayed responses and missed requests. Recent advances in speech-to-text, text-to-speech and large language models make it possible to build voice agents that can answer inbound calls, hold a structured conversation, collect the information needed to handle an issue, and hand it over to the support team in a consistent format.
Scope is adapted to 15 credits (bachelor) or 30 credits (master).
About Knightec Group Sweden
Who are we looking for?
- An Exciting Journey with Knightec Group Semcon and Knightec have joined forces as Knightec Group.
- Together, we are Northern Europe’s leading strategic partner in product and digital service development.
- With a unique combination of cross-functional expertise and a holistic business understanding, we help our clients realize their strategies – from idea to complete solution.
Responsibilities
- This thesis aims to design, implement and evaluate a voice AI agent for structured support intake, and to investigate how reliable and consistent this approach is compared to today’s alternatives.
- Project Description In this thesis, design and implement a voice AI agent that receives inbound customer phone calls in Swedish or English, guides the caller through a structured intake dialogue, handles follow-up questions within a controlled scope, generates a structured summary of the call, and delivers it into the existing support workflow (initially a shared mailbox).
- Building the prototype is required, but the contribution is the evaluation: how complete and consistent is the collected data, how faithful are the summaries, and how well current speech pipelines perform for Swedish and English support conversations.
- The prototype will be evaluated with simulated calls.
- LLM-driven with structured output) affect completeness and consistency of collected intake data?
- How accurate and faithful are automatically generated call summaries compared to the transcript What transcription accuracy and latency do available STT/TTS/LLM pipelines achieve for Swedish and English support calls, using the same intake flow in both languages?
- What architecture and integration points are needed to fit such an agent into an existing support workflow?
Requirements
- The agent itself is a means, not the end.
- Existing telephony, STT, LLM and TTS platforms/APIs should be used; building components from scratch is not expected.
- Bachelor/Master of Science in Computer Science/Engineering Graduation in June 2027 Fluency in Swedish, both written and spoken Fluency in English, both written and spoken Purpose Bachelor students typically focus on one question; master students take a broader comparative evaluation.
Preferred
- Compare design alternatives such as a scripted intake flow versus an LLM-driven flow with structured output.
In this thesis, investigate these questions
How do different dialogue control strategies (scripted flow vs.
Practical Information This is a thesis position, located at our office in Sundsvall . Start date January or March 2027. Please submit your application as soon as possible, but no later than 2026-11-30 . If you have any questions, you are welcome to contact [email protected] . Note that due to GDPR, we only accept applications through our careers page.
