Pi Pharma Intelligence Increases Database Speed and Size While Reducing Time to Resolution with AWS and cirrusgo

Pi Pharma Intelligence provides pre-market insights to stakeholders, such as researchers and analysts in the pharmaceutical industry, primarily serving the Middle East and North Africa (MENA) region. Its flagship service, Pi Insight, is a subscription-based platform built on AWS that allows customers to search for in-depth information about a drug before that drug is launched to the public. To provide this, Pi Pharma Intelligence gathers data from a variety of sources—patent submittals, clinical trials, academic research, and pricing information, among other things—to identify market risks, barriers, and opportunities. Customers can search for, and generate reports on, a host of different attributes, such as molecule types, price points, and dosages, using Pi Insight’s search interface.

As the company grew, so did its user base and the amount of data Pi Insight had to store and process, leading to slowed search times. Additionally, gathering, preparing, and reporting all this data came with many roadblocks, such as transforming non-uniform data formats. Because of the system’s centralized and expansive infrastructure, which included more than 800 lines of code, the Pi Pharma Intelligence team could spend up to an entire day resolving issues and finding their root causes. These challenges made entering new geographical markets and broadening datasets a daunting goal.

“The platform couldn't handle all the computation required for a search about a drug,” said Nusaibah Al-Jaloudi, founder and CEO of Pi Pharma Intelligence. “We didn’t want the system’s architecture to jeopardize the customer experience, so we decided to seek out expertise.” As a member of a startup accelerator, Pi Pharma Intelligence knew it had to rework its data pipeline if it wanted to expand internationally. That’s when the accelerator’s technical director recommended cirrusgo, an AWS Advanced Tier Partner located in the MENA region that provides custom, cloud-based IT solutions.

What sets Pi Pharma Intelligence apart from the competition is its proprietary poly-hierarchal database, which allows the company to relate a large amount of data points to each other in multiple ways. In its redesign, Pi Pharma Intelligence wanted to keep this structure. cirrusgo had to find a way to build additional data links and pipelines in a way that provided the scalability and performance the new structure needed—now and in the future.

cirrusgo worked with Pi Pharma Intelligence to deeply understand the business and the dependencies of the data to create proofs of concept. After testing solutions and finding the AWS services that fit best, cirrusgo built the new system from start to finish around Pi Pharma Intelligence’s poly-hierarchical data system. Pi Pharma Intelligence uses Amazon Relational Database Service (Amazon RDS) to generate reports and store them in a data lake built on Amazon Simple Storage Service (Amazon S3). This makes it possible for users to retrieve content with a single search. cirrusgo also helped Pi Pharma Intelligence develop an in-house CMS, which decoupled the infrastructure, allowing employees and customers to operate in separate environments. This separation not only improves performance by preventing slowdowns during searches but also facilitates maintenance, debugging, and updates to the CMS without impacting the customer experience.

To enhance the system's functionality, the teams implemented Amazon OpenSearch Service, which provides secure, real-time search, monitoring, and analysis of business and operational data as the primary database for insights. Amazon API Gateway enables the system to crawl the data for queries, further speeding up the search process. AWS Lambda and AWS Glue automates data loading from Amazon S3 to Amazon OpenSearch Service.

cirrusgo also helped Pi Pharma Intelligence ensure performance consistency by improving issue resolution. Amazon CloudWatch metrics, alarms, and logs address monitoring and observability, providing the teams with real-time analysis and incident troubleshooting. They can also visualize all the monitoring data through dashboards. “We can now guarantee that any data growth or system updates will go unnoticed by the customer,” said Nusaibah Al-Jaloudi. “This gives us all the freedom and flexibility to solve any problems we face.”

With the new AWS-powered data pipeline, Pi Pharma Intelligence sees greatly improved search speeds and reduced time spent generating reports—even while scaling its data. Before the redesign, a search could take several minutes; now, it takes milliseconds.

“We can make our data more complex and accommodate more attributes,” said Laith Al-Jaloudi, co-founder and technical manager at Pi Pharma Intelligence. “Our database is 10 times larger than it was a few years ago and we can accommodate data that is 179 times larger, so we’re providing more insights to customers.”

The new, decoupled structure that isolates components of the data pipeline from each other, along with robust AWS monitoring and observability tools, makes it easier for Pi Pharma Intelligence to manage potential performance issues without disrupting customer-facing interfaces. What formerly took the Pi Pharma Intelligence team an entire day to resolve now takes 10 minutes—a 98 percent reduction.

Presently, Pi Pharma Intelligence is collaborating with cirrusgo to use artificial intelligence (AI) tools, such as Amazon Bedrock, Amazon SageMaker, and Amazon Q Business, to manage differently formatted data, predict future events, analyze risk, assess opportunities, and drive data standardization. Pi Pharma Intelligence is also exploring how generative AI tools can help them build features such as engaging text summaries on industry developments or add value to reports. “The consistent performance that cirrusgo helped us achieve with AWS provides a great customer experience, and sets us up to grow internationally,” Laith Al-Jaloudi said.