The News: On January 24, Amazon posted a blog describing how the company is using generative AI to automate processes and improve customer experiences for Amazon Pharmacy. “Today, Amazon is using generative AI to make the pharmacy experience even better for our clinical teams and customers,” said Alexandre Alves, senior principal engineer at Amazon Pharmacy. “We saw an opportunity to rethink information flows that was both exciting and transformational.”
Here are the key details:
- Amazon Pharmacy is a full-service pharmacy in the Amazon.com store. Customers can use it to purchase medications prescribed by their doctor and have them delivered. Prime members have free, two-day delivery, the Prime prescription savings benefit, and RxPass, which offers access to unlimited eligible prescription medications for $5 per month. All Amazon Pharmacy customers have 24/7 access to a pharmacist to discuss questions about their medications.
- Alves, who is trained as a computer scientist and has pre-generative AI experience working with deep learning models to help streamline supply chain, led the initiatives.
- To power the generative AI capabilities, Alves and his team used multiple, pre-trained models from Amazon Bedrock and Amazon SageMaker.
- The team is using generative AI to address several issues, including how to:
- Streamline manual elements to digital prescription filling
- Forecast medication demand to streamline and speed delivery
- Enable clinical and customer care teams to answer medication questions faster
- Deliver medications to users more efficiently
- Digital prescriptions contain confusing or inconsistent language, particularly in the directions for the user. Workers sift through prescriptions by hand to edit, clarify, and confirm the data, which slows the process of filling prescriptions.
Amazon solved this challenge by running the original “unstructured” data through a generative AI model that uses a process called “named entity recognition” to create a structure for the text, using categories such as “dose” and “frequency.” From there, it helps Amazon Pharmacy clinical staff fill prescriptions and provide clear instructions for patients. Final prescriptions are still reviewed by a pharmacist to check the AI.
By combining a generative AI approach with the expertise of a pharmacist, Amazon has increased order processing speed by 90% and reduced the rate of human error.
- Forecasting medication demand helps Amazon stock the right medications in each shipping location, ready to be dispensed when a prescription comes in.
The team is using generative AI to synthesize data to test different stocking scenarios, both for how much to send to the shipping locations and for determining dispense methods at the sites.
- It is difficult in many instances to decide to fill orders individually or in batch. “By improving our ability to ingest data and interpret context, generative AI can help us improve these predictions and batching decisions, and help customers get their prescriptions more quickly,” Alves said. “If you don’t need a refill for a month, that request can probably wait until we can batch it. If you need a medication urgently, we’re going to bump it to the front of the line.”
Amazon Pharmacy: A Best In Class Case Study for Generative AI
Analyst Take: Generative AI has overrun the planet, in concept, theory, ideas. The fact is the practical use of generative AI is nascent. Amazon’s Pharmacy case study is an important one for the market that reveals lessons enterprises should learn and a blueprint they should follow in their journeys to unlocking the value of generative AI. Here is why.
Amazon Culture Plays a Part
Amazon Web Services (AWS) was born out of Amazon’s desire to explore how to better host the Amazon marketplace. There are two elements of Amazon culture that play into this generative AI use case story: 1) Amazon typically approaches innovation and opportunities because a customer presents it as a challenge. 2) Amazon’s many divisions are frequently “the customer.” In this case, Amazon Pharmacy has a natural connection and path to experimenting with the AI experts and platforms within AWS.
Amazon Starts with Problem-Solving and Expertise
Note the approach Amazon took: the team started with a process or operation that needed improvement. The company started with the business/operation problem first, then explored whether the technology could help solve it. This approach is fundamental, and the most critical first step to being successful with generative AI. Second, Amazon Pharmacy was able to move swiftly in part because the team leader, Alves, has AI experience. Not only does he have AI experience; he has it in areas most relevant to Amazon Pharmacy—supply chain management. The companies that will operationalize AI the earliest are companies that have been invested in AI for more than 2 or 3 years. These companies already understand the lifecycle management of AI—the risks, the need for data, etc. Many companies do not have the luxury of having an experienced AI technologist on staff and consequently, they are learning as they go. How do companies overcome that? Leverage vendor partners or consultants who do have AI and maybe even domain expertise.
Amazon Harnesses the Good of LLMs and Avoids the Bad
“Use LLMs as a reasoning engine to process information, rather than using it as a source of memorized information.”—Andrew Ng
This is my favorite AI quote. Note in Amazon’s use case, generative AI was not unleashed in text generation or conversational AI agents. Rather, Amazon channeled generative AI as Ng suggests—as a reasoning engine to process information. As large language models (LLMs) continue to evolve, some of their challenges might get solved. In the meantime, can your company take the risk LLMs present when they are used as Ng’s says, as the source of memorized information? I suggest no.
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Other Insights from The Futurum Group:
Amazon Q: What It Means for AWS’s AI Business
Amazon SageMaker HyperPod Claims 40% Reduction in AI Training Time
Guardrails for Amazon Bedrock Show AWS Gets Generative AI
AWS, Microsoft, and Google Cloud: Tying Up LLMs
Author Information
Based in Tampa, Florida, Mark is a veteran market research analyst with 25 years of experience interpreting technology business and holds a Bachelor of Science from the University of Florida.
