AI Development – Generative AI Assistant

Project Duration: 1 year Customer Company Size: 150 employees Industry: NGO focused on social impact YEAR: 2023-2024
Project Objective

The project’s mission was to create a Generative AI Assistant, an innovative AI-powered application to facilitate the crafting and dissemination of inclusive policies. The application utilized cutting-edge AI technology to offer intuitive interaction, striving to set a benchmark in digital accessibility and inclusivity.

Approach/Methodology

Employing a cloud-first approach, the project integrated advanced AI models within a browser-based application, underpinned by Microsoft Azure’s robust cloud and AI services. The project was marked by a comprehensive lifecycle, from initial development to meticulous testing, ensuring adherence to the Web Content Accessibility Guidelines (WCAG) and proactive prevention of representational harms.

Project Results

The Generative AI Assistant culminated as:

  1. An accessible, browser-friendly interface for user engagement.
  2. A testament to the potential of Azure SQL and Blob Storage in handling structured and unstructured data.
  3. A showcase of Azure Cognitive Search’s indexing process, coupled with Azure Logic Apps for data integration.
  4. A multilingual support system, extending its reach.
  5. A benchmark in quality, security, and accessibility, validated through rigorous testing.

Deliverables included a refined chatbot, comprehensive documentation, and user training, ultimately contributing to the foundation’s goals of fostering an inclusive society.

AI-Powered Recruitment – Jethire.ai

Project Duration: 2 years Customer Company Size: SMB’s, Large Corporates Industry: Cross-Industry YEAR: 2023-ongoing development

Project Objective

To revolutionize the recruitment process by integrating advanced artificial intelligence and large language models (LLMs), facilitating instant CV screening, unbiased candidate matching, and providing actionable recruitment insights, all while reducing the manual workload on recruitment teams by up to 80%.


Approach/Methodology

JetHire.ai employs LLMs to perform a semantic analysis of candidates’ resumes and job descriptions, going beyond keyword matching to appreciate context and nuanced skill sets. This includes:

  • Automated candidate screening and shortlisting based on precise skill matching.
  • Structured competency-focused interviews, enabling better candidate assessments.
  • Robust analytics dashboard for actionable insights and continuous recruitment optimization.
  • Multilingual proficiency to support global operations and cultural fit assessment.
  • Adherence to data privacy regulations, ensuring GDPR compliance.

Project Results

JetHire.ai’s AI-powered system has:

  • Reduced candidate screening time significantly, leading to a more efficient recruitment process.
  • Provided unbiased, data-backed candidate assessments, mitigating human biases in talent acquisition.
  • Ensured accurate skill-based candidate matching, with the system’s natural language processing capabilities recognizing the semantic relationships between various skills and job requirements.
  • Offered powerful analytics for data-driven recruitment, enabling companies to make more informed hiring decisions.
  • Demonstrated seamless implementation and scalability across different company sizes and industries.
  • Continuously learned and improved its recruitment algorithms, adapting to the evolving needs of the HR industry.

AI-Newsletter Management System

Project Duration: 25 days Customer Company Size: SMB’s Industry: Various YEAR: 2023

Project Objective

The objective was to develop an automated newsletter assistant that streamlines the creation, preparation, and distribution process of newsletters for businesses without manual intervention.

Approach/Methodology

The methodology involved integrating the assistant with Make.com’s webhook to handle specific tasks such as “DRAFT Newsletter Creation,” “Email Sending,” and “WORDPRESS Article Creation.” The assistant was designed to receive and process user inputs, ensuring the correct actions were triggered in the connected systems.

Project Results

The “Newsletter Assistant GPT” successfully provided a seamless newsletter management experience. It automated the drafting process by creating a newsletter draft when provided with the text. Furthermore, it prepared all emails for sending and executed the dispatch of these emails efficiently. For additional reach, it also automated the creation of WordPress articles with the content from the newsletters. The system has been commended for its efficiency and the time-saving benefits it offers to the company’s marketing team.

AI Development – GPT4 Website Assistant

Project Duration: 20 days Customer Company Size: SMB’s Industry: Technolgy & Tax YEAR: 2023

Project Objective

To develop a GPT AI Assistant that leverages Retrieval Augmented Generation (RAG) and Vector Database indexing to enhance user interaction on [Anonymized Company]’s website. The assistant is designed to provide real-time, accurate, and contextually relevant content retrieval, and dynamic response capabilities based on user inquiries.

Approach/Methodology

The project involved a multifaceted approach:

  • RAG Integration: Implementing Retrieval Augmented Generation to enrich the assistant’s responses with information pulled from an extensive index of the website’s content.
  • Vector Database Indexing: Employing advanced vector indexing techniques to semantically categorize and retrieve website data, allowing for nuanced understanding and matching of user queries.
  • User Experience Optimization: Designing the assistant’s interaction model to enable personalized and engaging conversations, continually adapting to user preferences and feedback.

Project Results

  • Enhanced User Interaction: The AI assistant significantly improved the way users engage with [Anonymized Company]’s website, offering an interactive and intelligent experience.
  • Accurate Information Retrieval: With RAG and vector database indexing, the assistant provided precise content related to user inquiries, demonstrating a deep understanding of the available information.
  • Dynamic Content Management: The assistant’s ability to generate and manage content dynamically resulted in a more robust and versatile website interaction platform.
  • Increased User Satisfaction: User engagement metrics showed increased satisfaction due to the AI assistant’s efficient and personalized responses.

AI Development – Arztbrief.ai

Project Duration: Ongoing development Customer Company Size: Various Industry: Healthtech YEAR: 2024

Project Objective

To develop an AI-powered solution for automating the creation of medical letters (Arztbriefe) in healthcare facilities, leveraging Natural Language Processing (NLP) technologies. This aims to significantly reduce the time medical professionals spend on administrative tasks, allowing for more focus on patient care.

Approach/Methodology

  1. Analysis of current workflows in medical report writing and identification of areas for process optimization.
  2. Utilization of Natural Language Processing (NLP) to interpret and structure medical data efficiently.
  3. Development of an AI model capable of generating comprehensive and accurate medical letters.
  4. Iterative testing and validation of the AI system in collaboration with healthcare professionals to ensure its efficacy and user-friendliness.

Project Results

Successful development and implementation of an AI system that can generate medical letters with minimal human intervention based on audio transcription. The system processes audio files and checks medical databases to create comprehensive and understandable medical letters, contributing to the efficiency of medical documentation and communication.