Amazon Bedrock Agents | Amazon Web Services

With Amazon Bedrock Agents, you can easily build and scale generative AI applications with security, privacy, and responsible AI.
This demo shows you how to use Amazon Bedrock Agents to configure specialized generative agents that seamlessly execute multi step tasks based on natural language input and your organization's data & policies. These managed agents play conductor, orchestrating interactions between FMs, API integrations, user conversations, and knowledge bases loaded with your data. As a developer, you can use the new trace capability to follow the chain of thought (CoT) reasoning that’s used as the plan is carried out. You can view the intermediate steps in the orchestration process and use this information to troubleshoot issues.
Learn more at: go.aws/3vC9RV8
Subscribe:
More AWS videos - bit.ly/2O3zS75
More AWS events videos - bit.ly/316g9t4
ABOUT AWS
Amazon Web Services (AWS) is the world’s most comprehensive and broadly adopted cloud platform, offering over 200 fully featured services from data centers globally. Millions of customers - including the fastest-growing startups, largest enterprises, and leading government agencies - are using AWS to lower costs, become more agile, and innovate faster.
#AWS #AmazonWebServices #CloudComputing #generativeai #generativeagents #promptengineering #taskorchestration #knowledgebases

[@AA-rd6nm]

Awesome work.

[@amazonwebservices]

Thank you! 😀

[@KratosKaChotaBhai]

The Claude model takes more than 10 seconds anyways to respond and then we added Bedrock agent which is again using Claude model to decide which action group to pick, and that will add again another 10 seconds to respond, i think as the use case is smaller in bedrock agent Amazon should have used or created there own SLM model to understand what the user want and select the action group instead of searching it in the LLM base

[@awssupport]

Hi there, thanks for the feedback! It's definitely appreciated. I've reached out to the Bedrock Support team regarding your input. One other thing I would like to share with you is our feedback article: go.aws/feedback. This article details how to provide the type of feedback you have given should you need to in the future. However, you can always reach out here and, we will do our very best to help. ^DC

[@equinozio22]

Can interact through appsync or has to be a rest api?

[@awssupport]

Hi, there. 👋 This doc walks you through developing Rest APIs with AppSync: go.aws/411E38v. 📃 If this doesn't quite answer your question, you can also reach out to our community of helpful engineers at re:Post: go.aws/aws-repost. 📝 ^ZP

[@equinozio22]

@awssupport that doesn't make any sense... Are you a bot?

[@awssupport]

Definitely not a bot! 🤖 Our scope for tech support is limited on this platform, which is why we sent the previous options. 🤝 ^RW

[@anishshrestha2378]

Cane we do response stream in agents?

[@awssupport]

This is an excellent question for our re:Post platform! 📮 Our cloud community of experts may be able to assist you with an answer or additional advice: go.aws/aws-repost. 🤝☁️ ^TE

[@michaelinzo]

Can we integrate it on discord or slack etc?

[@renatodex]

Omg is the API slow like that? The response in your test application took like 40 seconds to complete!
Can't believe anyone noticed that haha

[@awssupport]

Hi, thanks for your observations. 👀 Perhaps you would like to share this info with the appropriate team: go.aws/feedback. This article details how. We definitely value info from our customers. ^DC

[@canasdruid]

how can i train this model with company pdfs or excels?

[@awssupport]

Hi there! 👋 Our community of developers over at re:Post might be able to lend their expertise & guidance: go.aws/aws-repost. Additionally, we have these other ways to get help: go.aws/get-help. ⬅️ ^AK

[@rishirajkalita9883]

Fine tune or knowledge bases

[@canasdruid]

@@rishirajkalita9883i am currently doing that

[@dozo_ca]

Yeah I was thinking about that

[@jeffdeng1652]

@@dozo_ca just point the knowledge base at the company docs as a datasource and describe the data in the docs during setup and the agent will figure out when to use the data inside.

[@nickdemiceli4543]

Is it possible to leverage the knowledge base of my agent in Amazon bedrock, kind of as a guide for how the agent should respond. For example, I have a very specific use case and I only have around 30 examples of how it should respond, so I believe fine tuning would not work, but i am curious if I put 30 examples into the knowledge base, would it guide the agent better.

[@awssupport]

Great question, Nick. Our community of experts at re:Post may have technical insight to offer on this: go.aws/aws-repost. ✅💻 ^MO

[@rptechplex]

You may be thinking of agentic ai with RL. This is about AI agents working within action boundaries.