Artificial intelligence (AI) has begun transforming industries. As an innovative technology builder, out engineering team at MOHARA have been super keen to get the opportunity to build PoC internally for challenges we feel and also build out tools for our portfolio and partner.
This case study article explores two notable projects in two verticals: Healthcare and Automotive. I have omitted the names of the companies as both of these companies as of the time of writing are in stealth mode.
Healthcare: Revolutionising Patient Engagement
Challenge:
A significant portion of patients, approximately 70%, do not fully understand their treatment plans, leading to suboptimal health outcomes. This startup sought to address this issue by creating automated, interactive video content that features the caregiver’s digital twin. This content aims to deliver more effective call-to-action instructional videos in multiple languages, improving patient comprehension and engagement.
Solution:
MOHARA partnered to design and develop a platform that enables users to brand, customise, configure, and distribute these digital twin videos. The platform also allows users to create, edit, and track videos and their performance, ensuring a seamless and impactful patient experience.
Tech Stack:
React with Next.js: For building the front-end interface.
Django REST API: To serve the backend.
Python Lambdas: For asynchronous processing.
AWS Services: For hosting and infrastructure.
GenAI Components: To integrate advanced AI capabilities.
Terraform Pipelines: For infrastructure management.
Impact:
This innovative solution has empowered healthcare providers to deliver personalised, comprehensible, and accessible treatment plans, significantly improving patient engagement and adherence to medical advice.
Automotive: Enhancing BOM (bill of materials) Validation with AI
Challenge:
In the vehicle manufacturing industry, validating the Bill of Materials (BOM) is a critical yet complex task. A BOM is a detailed list of all components, parts, and materials required to assemble a vehicle. Ensuring its accuracy is vital for maintaining quality, efficiency, and compliance. However, manual validation is prone to errors and can be costly.
Solution:
MOHARA collaborated with an established data consultancy to develop an AI-powered product that assists BOM validation engineers. Utilising large language models (LLMs) like ChatGPT, this product helps engineers identify issues earlier in the process, saving time and reducing costs.
Tech Stack:
Django full stack: For the application development.
Open AI API: To leverage advanced language models.
Langchain: For integrating and optimising AI workflows.
Features:
Prompt Engineering: Creating and testing logical rules against the BOM dataset.
Summarisation and Reporting: Efficiently summarising and reporting outcomes.
Retrieval Augmented Generation (RAG): Retrieving relevant context from proprietary knowledge bases to validate parts and correct errors.
Model Fine Tuning: Enhancing model accuracy using synthetic data.
Code/Query Generation: Generating SQL queries from natural language inputs.
AI Vision: Identifying vehicle parts in images and cross-referencing them with the dataset.
Guardrails: Ensuring outputs are relevant and helpful.
Impact:
By automating the BOM validation process, the consultancy has significantly reduced errors and streamlined their workflow, leading to improved efficiency and substantial cost savings.
Additional AI Projects by MOHARA
MOHARA’s expertise extends beyond healthcare and automotive industries. The firm has spearheaded various AI proof-of-concept (PoC) projects across different sectors:
Construction Job Marketplace:
Transforming informal job advertisement texts into structured data using a lightweight PoC with the T3 stack and Langchain.
ED&I Consultancy:
Productising an audit process with survey platforms, speech-to-text technology, and AI for qualitative to quantitative assessments and recommendations.
Recruitment Agency:
Enhancing interaction with existing knowledge bases through a chatbot built using Azure services and integrated with Slack.
Property Management Service:
Developing an automated data extraction and document classification tool using Google Document AI and Flutterflow web app.
Tender Copilot:
Creating a tender application writing assistant that references past bids and case studies using RAG and integrating Anthropic models into a low-code web app.
MOHARA’s commitment to pioneering AI solutions has enabled their partners to overcome industry-specific challenges with innovative, effective, and scalable technologies. From healthcare to automotive and beyond, MOHARA’s projects exemplify the transformative potential of AI, setting new standards for efficiency, accuracy, and user engagement. As industries continue to evolve, MOHARA remains a trusted partner in leveraging AI to drive progress and achieve remarkable outcomes.
If you are interested in building an AI enabled PoC please feel free to reach out to us at [email protected]
