Case Studies

How We've Solved Challenges for Our Clients

Medical Data Broker

Client Overview

The client was receiving millions of handwritten medical prescriptions in different regional languages as PDFs. They needed to anonymize this data to make it HIPAA compliant and put it in a standard format.

Challenge
  1. Anonymize and standardize medical prescription data.
  2. Perform ad-hoc queries to look at global trends.
Solution
  1. A pipeline was created using Optical Character Recognition (OCR).
  2. A custom Gen AI bot translated all prescriptions to English.
  3. The AI model generated insights into different medication prescriptions per region and sold this data as a subscription to pharmaceutical firms.
Results and Impact
  1. Access to real-time data and historical trends.
  2. LLM 'function calling' used for ad-hoc queries across countries, age groups, morbidities, and prescriptions.
  3. Data-driven approach allowed focus on regions with the biggest sales opportunities.

Medical Supplement Interaction

Client Overview

The client needed a system to allow customers to check for allergic reactions or problematic medical interactions before buying supplements.

Challenge
  1. Some supplements caused customer problems and side effects due to allergic reactions, affecting sales and reputation.
Solution
  1. A comprehensive database was developed using scraping and LLM agents.
  2. Customers scan supplement labels using OCR before buying.
  3. Possible dangerous interactions are identified immediately before purchase.
Results and Impact
  1. Improved customer satisfaction and sales.
  2. Prevention of possible allergic reactions.
  3. Provided guidelines on which supplements to buy based on goals and current supplement intake.

Networking Application

Client Overview

The client, a startup, wanted to build an AI-based personal networking concierge.

Challenge
  1. Assist users in unlocking the full potential of their personal network.
Solution
  1. Developed an AI Chat Bot that allows users to describe the type of person they are looking for in natural language.
  2. Utilized LLM graph databases, speech-to-text, and text-to-speech technologies.
  3. Enriched contact data by scraping LinkedIn and exploring links between contacts and companies.
Results and Impact
  1. Enabled linkages between contacts and locations.
  2. Provided agents for real-time web searches.
  3. The customer is in an invite-only paid beta phase with a strong waiting list of potential customers.

Customer Service

Client Overview

The client wanted an AI Chat Bot for a comprehensive 24/7 customer engagement platform.

Challenge
  1. Provide a comprehensive AI-based customer engagement platform.
Solution
  1. Developed an AI Chat Bot that allowed users on different engagement platforms to communicate and get comprehensive answers to their queries.
Results and Impact
  1. Relieved staff from mundane, repetitive work.
  2. Increased staff productivity and general happiness with their work.

Veterinarian Efficiency System

Client Overview

The client, providing veterinary clinical services, wanted a platform to generate insights from clinical data and improve office efficiency.

Challenge
  1. Generate comprehensive insights from clinical data and make the office more efficient.
Solution
  1. Developed a Generative AI Chat platform trained on the customer’s clinical records using OCR.
  2. Provided insights based on simple prompts.
  3. Automated the filling out of customer record paperwork.
Results and Impact
  • Continuous insights from data improved operational efficiency and customer satisfaction.

Medical Claim Processing

Client Overview

The client needed an efficient way to process medical claims using existing information from medical facilities.

Challenge
  1. Efficiently process medical claims from unstructured textual data like doctors' notes, customer personal data, and payment information.
Solution
  1. Developed a system using a graph database to ingest unstructured textual data.
Results and Impact
  1. Significant improvements in processing efficiencies.
  2. Provided explainable decisions for claim acceptance or rejection.
See Details

E-Waste Application

Client Overview

The client wanted to identify precious metals from electronic waste material using a mobile format.

Challenge
  1. Identify precious metals from PCB boards and other waste materials in a mobile format.
Solution
  1. Used vision computing to identify and classify precious metals from over 100 categories.
  2. Enabled mobile phone application use through a cloud-based system.
Results and Impact
  1. Real-time, on-site size and dollar value estimates of expected metals to be recovered.
  2. Gained significant advantages in sales transactions through real-time data.