Overview

Project scope

Perelyn partnered with Stiegeler to accelerate customer support and enhance customer satisfaction through an AI-powered chatbot. The chatbot will leverage a comprehensive technical knowledge base and live data to provide precise answers to internal support teams.

Key Objectives & Deliverables

1

Implement a customised chatbot for querying the company's knowledge resources.

2

Enable a faster and improved customer support by internal operators.

3

Adding an external-facing customer chatbot interface on the Stiegeler website.

Stiegeler is a Germany-based telecommunications provider delivering high-speed internet, telephone, and TV services via fiber optic and DSL. They prioritise reliable connectivity for both residential and business customers, with a focus on rural areas, ensuring consistent access to essential communication technologies. Learn more about Stiegeler here.

Telecom industry & AI

The projected value of AI in the telecommunications industry is expected to reach USD 58.74 billion by 2032. AI is transforming our client’s industry by optimising network performance, enhancing customer service with intelligent chatbots, and strengthening security through advanced fraud detection. It is also driving innovation in 5G, edge computing, and IoT, reshaping how companies deliver smarter connectivity, personalised services, and more efficient operations.

Stiegeler’s goal is to maintain high-quality customer support 24/7. The challenge lies in achieving this seamlessly, as their outsourced after-hours provider lacks the deep company and technical expertise required to match the high quality and efficiency of their in-house team. To maintain excellent service around the clock, Stiegeler’s customer support team needs easy access to comprehensive knowledge and resources, ensuring that customer inquiries are addressed promptly and accurately, regardless of who is handling the support.

“At Stiegeler, we believe in empowering our team with consistent and reliable information across the board. That’s why we entrusted Perelyn with the development of a chatbot that ensures all employees are on the same page, especially in delivering accurate and timely responses in customer support. We sincerely thank Perelyn for the excellent collaboration and the innovative solution they’ve provided."

— Felix Stiegeler, Managing Director Stiegeler

Our approach

Perelyn hat für Stiegeler eine vollständige, AWS-basierte KI-Chatbot-Lösung entwickelt und implementiert. Zu den Funktionen gehören eine multimodale RAG-Konversation, Token-für-Token-Streaming mit der neuesten Generation von LLMs, eine benutzerfreundliche Oberfläche und eine Schnittstelle zu internen Unternehmensdaten. Alle Elemente wurden speziell auf Stiegelers einzigartige Anforderungen und Bedürfnisse zugeschnitten, um ihre Use-Case-Anforderungen optimal zu erfüllen.

Industry perspective

KI-gestützte Chatbots bieten Telekommunikationsunternehmen 24/7-Support, sofortige Beantwortung von Anfragen und Skalierbarkeit, wodurch Betriebskosten gesenkt und die Kundenzufriedenheit gesteigert werden. Sie liefern personalisierten, konsistenten Service, verfügen über mehrsprachige Fähigkeiten und generieren wertvolle Datenanalysen. Gleichzeitig integrieren sie sich nahtlos mit menschlichen Agenten, um komplexere Anliegen zu bearbeiten.

LLM-gestützte Chatbots können Unternehmen bis zu 30 % der Kosten im Kundenservice einsparen.
Dennoch nutzen lediglich 20 % der Unternehmen diese Technologie.

Project phases

Phase 1
Readiness & Prototyping

Definition der Chatbot-Funktionen: Customising document and API querying, usage profiles, and usage analysis.

Entwicklung eines Minimalprototyps: Showcasing the capabilities of document and API querying.

Implementierung eines benutzerfreundlichen internen Frontends: Building a modern React-based web frontend.

Phase 2
Implementation & Deployment

Bereitstellung auf AWS: Hosting the full-stack solution on AWS.

Optimierung der Antwortgenerierung: Enhancing quality through improved document retrieval and prompt fine-tuning.

Interne Produktionsfreigabe: Rolling out the solution across different departments within the company.

Stage 3
Improvement & Rollout

Kontinuierliche Verbesserung: Leveraging human feedback on chatbot results and the overall solution to drive enhancements.

Externe Einführung: Refining Chatbot responses and the frontend for deployment to external customers.

Phase 4
Handover

Umfassende Schulung: Conducting training sessions to ensure seamless adoption.

Wissensweitergabe: Delivering detailed documentation and hands-on training for future maintenance and updates.

Langfristige Partnerschaft: Equipping Stiegeler with the tools and guidance for ongoing improvements, while ensuring Perelyn’s expertise is always available.

Phase 1
Vorbereitung & Prototyping

Definition der Chatbot-Funktionen: Anpassung von Dokument- und API-Abfragen, Erstellung von Nutzungsprofilen und Analyse der Nutzungsdaten.

Entwicklung eines Minimalprototyps: Demonstration der Fähigkeiten zur Dokument- und API-Abfrage.

Implementierung eines benutzerfreundlichen internen Frontends: Aufbau eines modernen, React-basierten Web-Frontends.

Phase 2
Implementierung & Bereitstellung

Bereitstellung auf AWS: Hosting der Full-Stack-Lösung in der AWS-Cloud.

Optimierung der Antwortgenerierung: Verbesserung der Qualität durch präzisere Dokumentenabrufe und Feinabstimmung der Prompts.

Interne Produktionsfreigabe: Einführung der Lösung in verschiedenen Abteilungen des Unternehmens.

Phase 3
Verbesserung & Rollout

Kontinuierliche Verbesserung: Nutzung von Feedback der Mitarbeiter zu Chatbot-Ergebnissen und der Gesamtlösung, um Optimierungen voranzutreiben.

Externe Einführung: Verfeinerung der Chatbot-Antworten und des Frontends für den Einsatz bei externen Kunden.

Phase 4
Übergabe

Umfassende Schulung: Durchführung von Trainings, um eine nahtlose Einführung der Lösung zu gewährleisten.

Wissensweitergabe: Bereitstellung detaillierter Dokumentationen und praxisnaher Schulungen für die zukünftige Wartung und Weiterentwicklung.

Langfristige Partnerschaft: Ausstattung von Stiegeler mit den notwendigen Werkzeugen und Anleitungen für kontinuierliche Verbesserungen, bei gleichzeitiger Verfügbarkeit der Expertise von Perelyn.

Need your own chatbot?

Overview

The Stiegeler Chatbot was designed to assist in answering customer queries. Perelyn conceptualised and implemented a full-stack chatbot solution for Stiegeler within their AWS environment, with a special focus on customised document processing, storage and retrieval, as well as company API querying capabilities.

The customer asks what wifi routers Stiegeler has on offer that are compatible with fiber optics connections?

Hallo! Stiegeler Internet Services bietet eine breite Auswahl an WLAN-Routern an. Die folgenden Modelle sind mit Glasfaseranschlüssen kompatibel und können auf der Stiegeler-Website gekauft oder gemietet werden:

Fritz!Box 7690
Fritz!Box 7530 AX

Let me know if you require anything further!

Business impact & benefits

The chatbot solution enables Stiegeler to enhance customer support across their telephone and email channels, as well as their website. The results are clear: reduced turnaround times for customer queries and a higher success rate, leading to increased customer satisfaction.

Technical architecture & integration

Perelyn integrated advanced LLMs with modern methods, including text search, conversational RAG, and a dynamic front end, to deliver a smooth, user-friendly chatbot experience. The system was seamlessly integrated into the client's existing cloud infrastructure, with special emphasis placed on data protection considerations.

“We aimed to empower Stiegeler with a chatbot that transforms the customer experience through quick, reliable support, creating a solution that meets their needs today and scales for tomorrow.”

– Max Schattauer

More on the topic

Insights

Telecom sector: How AI is Changing the Game

Artificial intelligence optimises networks, improves customer service through personalised chatbots, and enables new data-driven business models. However, challenges such as data privacy must be addressed. AI offers great potential for the industry.

Key Tech

LLM Frameworks

AWS Bedrock, Amazon Comprehend, Google VertexAI, Azure Open AI, HuggingFace Transformers, LangChain, LangSmith, LLamaIndex

Graphdatabases

Anthropic, OpenAI, AWS Bedrock

Modern Frontend

React

Backend Services

Docker, REST APIs

Cloud

AWS, ECS, CloudFront, MongoDB

Programming languages

Python, TypeScript

Relevant insights

Generative AI
The next big leap in retail

Discover the impact of generative AI in retail. This technology unites departments, personalises content, and transforms customer experiences. Leading brands such as Coca-Cola and Walmart are already using their potential to optimise operations and drive innovation. Explore the future of retail with generative AI...

#LLM
#Generative AI
Sebastian Fetz
28.2.2024
Read
Generative AI
How do you use LLMs with internal company data?

This article explores how large language models such as ChatGPT can be integrated with proprietary data. The recent advent of large language models opens up numerous opportunities for companies, but connecting with proprietary data is a challenge in such a transformation. This article discusses the various approaches to solve this problem, such as fine-tuning and contextual learning. In addition, the associated challenges and risks are addressed.

#LLM
#Generative AI
Dominik Filipiak
19.2.2024
Read
Generative AI
Superhuman Assistant - LLMs as a Strategic Priority

In this article, our CEO Sebastian highlights the transformative power of large language models (LLMs) in a corporate context and provides guidance for their implementation. This is not just a dream of the future, but a strategic necessity that future-oriented companies are taking on today to drive groundbreaking innovations.

#LLM
#AI Strategy
#Generative AI
#LLMOps
Sebastian Fetz
10.1.2024
Read

Contact us

Name*
Email*
message*
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Help us improve our website. By “Accept” click, do you agree that we store a cookie on your device to analyse the use of the website. Read our Privacy statement for more information.
Something went wrong. Please try again.