AWS has launched “Agents for Amazon Bedrock”, which helps developers create generative AI applications that provide answers and complete tasks.
While most generative AI applications leverage large language models (LLMs) to produce information of one form or another, Agents for Amazon Bedrock can perform actions.
A possible use case of this technology might be train an app to make a hotel or restaurant booking on a user’s behalf instead of simply suggesting options for you to reserve yourself.
Yet, while AWS uses “agent” in the name, Agents for Amazon Bedrock is not an end-customer-facing technology. It does not deliver self-service experiences for a customer service use case(s). Rather, Agents for Amazon Bedrock can enable generative AI apps to provide self-service experiences.
As such, Amazon Bedrock is not a technology that will be used by customer service agents to serve end-customers. Instead, developers can lift the solution to create various applications that improve customer experiences.
Using the reservations example, developers could create use Agents for Bedrock to support a machine customer application that schedules bookings for customers.
That application may support Gartner’s prediction that a fifth of all customer contacts will be handled by machine customers before 2026.
Alongside other solutions, like Google Duplex, such innovation may make this estimate a reality.
Yet, Swami Sivasubramanian, Vice President of Data and Machine Learning at AWS, pinpoints further use cases. He told The Verge:
I believe this will supercharge developers who wanted an easier way to build agents and at the same time customise the data the models read.
“Building agents took so much time even with how advanced generative AI is now, but we’re making it so developers can access exactly the models they need.”
Indeed, Agents for Amazon Bedrock is a fully managed capability that enables generative AI applications to complete tasks in a few clicks without any manual code required.
It can orchestrate interactions between FMs, data sources, software applications, and users and execute APIs automatically.
Developers can quickly and easily integrate ‘agents’, speeding up the delivery of generative AI applications and saving themselves a lot of time.
More on Amazon Bedrock
Amazon’s cloud business, AWS, is creating a database of generative AI tools under the umbrella of ‘Bedrock’.
Launched in April this year, Bedrock hosts third parties generative AI applications on its cloud alongside its own.
This sets it apart from the likes of Google and Microsoft, which have chosen to build their models themselves.
Agents for Amazon Bedrock was launched at the AWS Summit in New York on Wednesday.
The new application adds to a growing list, including Cohere’s FM selection, Claude 2 from Anthropic, Stable Diffusion XL 1.0 from Stability AI, Amazon’s own Titan language, and more.
It has been developed with start-ups and enterprises in mind, allowing companies to use their own data to train the foundational models, such as large language or image-to-text models, as well as create additional apps to complete tasks.
Elsewhere, AWS announced the application of generative AI to healthcare via its new service, AWS Health Scribe, which automates clinical documentation and makes medical notes for diagnoses.
The cloud leader also released Entity Resolution, a solution that sweeps enterprise applications and connects data to surface new customer and business insights.