Request a Call
Spinner

Processing...

Business Overview

Our client, a global leader of providing digital solutions for pharmaceutical companies, delivers sophisticated complaint management software tailored to healthcare companies of all sizes. Their highly customizable quality management system (QMS) allows healthcare and life science organizations to conduct supplier quality control, digital content, document, and audit management, regulatory compliance, and other in-house operation optimization. However, as it was developed a long time ago, this software struggled to keep up with modern digital trends, posed operational and security risks, and significantly slowed business growth.

To address these risks, the client approached NIX. They required partial modernization, ongoing maintenance of current features, and, more importantly, the integration of advanced AI-driven functionalities, featuring generative AI and natural language processing to automate manual workflows and accelerate pharmaceutical companies’ customer complaint resolution.

Given our successful track record with the client and our deep expertise in AI, data science, and machine learning, NIX was engaged to augment their in-house development team and enhance their AI capabilities.

Challenge

The software’s complex legacy nature entailed inefficient complaint operations, user outflow, and consequently, the loss of leading market positions to competitors. However, our team managed to adapt the solution into modern architecture on time while keeping the core functionality of the running models.

Solution

Collaborating closely with a stakeholder and the associated director of the R&D department, our specialists translated their ideas into actionable technical solutions, additionally conducting investigations and developing MVPs and PoCs. The highly customizable nature of the software required simultaneous maintenance of various product versions while refining outdated parts and integrating new AI functionalities.

2_600x405@2x

We developed various third-party AI APIs, utilizing the Azure ecosystem, including Azure Cloud, Azure OpenAI, Azure Machine Learning, and Azure Functions. For back-end development we selected Azure Functions due to its serverless architecture and robust orchestration capabilities, ensuring efficient processing of multiple data sources. The choice of Azure OpenAI was driven by its on-demand pricing model, providing a cost-efficient and scalable solution for the client.

3_600x405@2x

To significantly simplify and accelerate the grouping of similar customer complaints, we developed a similarity solution, which leverages Facebook AI Similarity Search (FAISS) algorithms. This solution analyzes historical data and utilizes vector representations to automatically identify and cluster customer complaints based on semantic similarity, even if their phrasing differs. It also reduces manual effort, improves efficiency, and enhances data analysis for faster complaint resolution.

4_600x405@2x

Our specialists further capitalized on generative AI capabilities, prompt engineering, and retrieval augmented generation (RAG) when developing core features for the software. To make model training as efficient as possible, we used Azure Machine Learning GPU Compute Clusters as accelerators with Nvidia T4 GPUs, as well as Parameter-efficient Fine-tuning (PEFT) with LoRA, and Hugging Face’s Trainer API. Additionally, to ensure correct usage of RAG, we leveraged text-embedding-3-large and all-MiniLM-L6-v2 models.

5_600x405@2x

As a result, we executed seven core features that delivered substantial improvements to the client’s software:

  1. AI summarization tool: Automates the summarization of complex metadata, such as descriptions, product names, codes, dates, and PDF and DOCX files, enabling faster understanding and analysis for support teams.

  2. Multiple acceptance criteria field: Streamlines complaint validation by automatically assessing input against predefined criteria. This provides a “valid” flag, clear reasoning for the assessment, and actionable recommendations for improvement, reducing manual review time and ensuring data quality.

  3. Automated email template generation: Improves response times and consistency in customer communication. AI extracts key information (e.g., name, symptoms) from incoming complaints and seamlessly generates pre-defined email templates, providing ready-to-send responses in a two-step process.

  4. Severity-based issue classification: Conducts a robust risk assessment by providing AI-driven recommendations for complaint severity (e.g., critical, major, minor) based on historical data and event descriptions. This allows for quicker prioritization and allocation of resources, leading to more efficient complaint management.

  5. Audit report generation: Generates tailored audit reports by comparing user input against static regulatory guidance. Users can specify key topics and desired report structure, giving them control and flexibility while ensuring compliance.

  6. Document change summarization: Semantically analyzes documents, identifies and tags all changes, and generates a short, concise summary of the main modifications. This significantly reduces the time and effort required to review lengthy documents (e.g., 200–300 pages) from hours to mere seconds, improving efficiency and accuracy in critical document management.

  7. Manufacturer and user facility device experience (MAUDE) code classification model: Automates the critical task of categorizing medical device event descriptions into specific FDA codes, which significantly streamlines compliance and reduces manual classification errors. Leveraging GPT-4o for robust inference and an advanced text-embedding-3-large model, the feature employs custom algorithms to effectively manage data imbalances and achieve high accuracy.

6_600x405@2x

Outcome

Ultimately, the client received significantly improved software that automated the complaint resolution process and significantly increased customer platform adoption. By providing comprehensive technical expertise to support the client’s ideas, our specialist transformed them into tangible, successfully deployed AI-driven solutions. Furthermore, the project is still continuing, and our team is actively expanding functionality to further enhance the solution and deliver even greater value.

Team:

Team:

.NET Architect Data Scientist JS Developer
Tech stack:

Tech stack:

Azure AI Azure .NET Azure Functions React RAG FAISS

REQUEST A CONSULTATION

Contact us   

Relevant Case Studies

View all case studies

Population Health Platform: Development and Modernization

Healthcare

Success Story Population Health Platform: Development and Modernization image

Next-Gen eLearning Platform for Medical Schools

Healthcare

Education

Success Story Next-Gen eLearning Platform for Medical Schools image

Efficient Patient Monitoring Ecosystem

Nonprofit & Charity

Healthcare

Success Story Efficient Patient Monitoring Ecosystem image

Thomson Reuters Platform

Pharmaceutical

Healthcare

Success Story Thomson Reuters Platform image

Cloud-based Solutions to Enhance HR and Behavioral Health Platforms

Healthcare

Success Story Cloud-based Solutions to Enhance HR and Behavioral Health Platforms image

AI-powered System: Cybersecurity Report Generation and Risk Mitigation

Healthcare

Success Story AI-powered System: Cybersecurity Report Generation and Risk Mitigation image
01

Contact Us

Accessibility Adjustments
Adjust Background Colors
Adjust Text Colors