Artificial intelligence took center stage at the AWS Summit in New York last week as the public cloud giant unveiled and showcased the capabilities available to partners and customers in the cloud.
“Why is generative AI technology just getting all the attention right now?” asked Swami Sivasubramanian, vice president of databases, analytics and machine learning at AWS, in his keynote address. “That is because this technology has reached its tipping point – the convergence of technological progress and what it can accomplish today.”
ChatGPT has certainly sparked a flame of creativity among businesses and individuals who are finding innovative use cases for such technologies every day.
The following forces have brought us to this moment, according to Sivasubramanian: the massive proliferation of data, the availability of extremely scalable compute infrastructure, and the machine learning technology advancements.
The result is something called foundational models. A foundational model is a large machine learning model that is trained on a vast quantity of data at scale. These can then be fine-tuned to a user’s specific needs.
AWS Summit NYC Announcements
With that in mind, here’s a high level view of some of the announcements the public cloud company made at its AWS Summit NYC last week:
- A preview of agents for Amazon Bedrock. In April AWS announced Amazon Bedrock and Amazon Titan models. Amazon Bedrock is currently available in preview and is a managed service that makes foundation models (such as those available from Amazon and some top AI startups) available via an API. Agents for Amazon Bedrock enables developers to create fully managed agents in a few clicks, according to AWS. Sivasubramanian provided the example of creating a complex chatbot and the many calls such a tool needs to make to many different systems. These Agents simplify the process of creating such a chatbot. Fully managed agents provide support for monitoring, encryption, user permissions, and API invocation management without writing custom code.
- AWS Entity Resolution: match and link related records from multiple applications and data stores. This machine learning-powered service offers advanced matching techniques such as rule-based matching and machine learning models to help accurately link related sets of customer information, product codes or business data codes. Organizations can leverage this to create a unified view of customer interactions.
- AWS Healthscribe preview. This new HIPAA-eligible services is designed to help healthcare software vendors to build clinical applications that automatically generate preliminary clinical notes by analyzing patient-clinician conversations.
- A new technical deep dive course: Generative AI Foundations on AWS. This 8-hour course is a curated playlist of the top resources, concepts and guidance to get up to speed on foundation models, especially those that unlock generative AI capabilities in customer/partner data science and machine learning projects.
- AI, generative AI, and machine learning guides for customers/partners to plan their AI journeys. AWS has published two new guides – the AWS Cloud Adoption Framework for Artificial Intelligence, Machine Learning, and Generative AI and the Getting Started Resource Center machine learning decision guide.
IBM’s Partnership with AWS on AI
Jay Limburn, IBM Distinguished Engineer and Vice President of Product Management at IBM, told ChannelE2E that foundational models allow the use of much larger sets of data. If you are using a foundational model you don’t need to go out and find and train the model. Instead you tune the model with the data that is specific to your organization and its use cases.
“This lets us be able to think about creating models that are trained on relevant enterprise data,” he said.
Limburn noted that IBM has huge volumes of enterprise data. IBM has also gone through its own process to scan for obscenity, profanity and copyrights, thus eliminating some of the risks that are concerning to large enterprises.
“ChatGPT has been banned by many organizations,” he said.
What’s more, the IBM implementation tackles another crucial enterprise need – providing explainability of the results. Often, when a model produces results, business executives and AI ethicists will ask how the “black box” of the model arrived at that answer. “Explainability” is a function in some systems that enables a look inside the model and how it came up with a particular output.
Limburn said that IBM has 10,000 consultants trained to use the company’s technology on AWS. The company is also working on technology, such as AutoAI, to make it easy for business users to leverage the power of AI.
AWS Marketplace Partner Perspective: Veriff
Brian Krause, Vice President, Global Revenue at Veriff, a company that provides digital identification technology, is in the midst of expanding its go-to-market program to do more with AWS Marketplace. The company has seen fast growth due to the increased demand for digital identity services that was sparked by the pandemic. Krause notes that Veriff runs on AWS and leverages some of the cloud providers services within the Veriff platform. Veriff has been in the AWS partner program since January of 2022. The company is in the process of launching on AWS Marketplace now. The move follows on Veriff’s B round of funding in March 2021, with much of the capital invested in go-to-market teams, Krause said. Veriff is currently doing API work now and expects to be on AWS Marketplace within a few weeks.
AWS Marketplace Update
Chris Casey, Director and General Manager, Industry Technology Partnerships at AWS told ChannelE2E that the AWS Marketplace currently has more than 10,000 listings for more than 3,000 different software sellers, providing a huge breadth of solutions. Most recently he has seen a build out of more industry-specific software solutions on the Marketplace, something that may be of interest to managed service providers who offer a vertical market specialization.
For customers and partners interested in generative AI, Casey noted that C3.ai has a standalone generative AI application listed on the AWS Marketplace. The company is one of the preview partners for Amazon Bedrock as well. This allows a customer to combine their internal data with the information that is trained on the large language models (such as ChatGPT). This happens in a private instance, so the organization’s data remains private and proprietary.