How To: Amazon Q - Step by Step Guide

[@nhtna4706]

love to see the end result, where you mentioned that you would be building something for your client in dec. am assuming you may have already built it,. looking forward for a live demo

[@Flowprodev-yp1tg]

It's mostly that a company will use Q on internal documents -- so this could be the knowledge base with all the documents they have about a certain project. If then a new member joins the project they can ask Q all kinds of questions about the project, specifications, etc. It's very powerful not only for new project members but also saves a lot of time organizing and gathering information about what was decided or discussed in the past. I'll see if I can show a demo of one but most of the times clients do it with confidential information unfortunately.

[@Flowprodev-yp1tg]

Use this link to create an AWS account: docs.aws.amazon.com/SetUp/latest/UserGuide/setup-AWSsignup.html

[@ArjunS-k1f]

I really enjoyed the video! Is there an API available to connect to the service created in Amazon Q, so that I can use it in my local projects to get responses?

[@cre8love158]

i am looking in the aws pricing calculator and i cant find Amazon Q. Is this a free service from Amazon?

[@Flowprodev-yp1tg]

The service is still in preview and they haven't updated the calculator yet. Here is the pricing page: aws.amazon.com/q/pricing/

[@ramya12132]

Instead of deploying the application Is there a method to obtain an API endpoint for Amazon Q application that can be directly utilized externally?( Ex: Postman ) and Thanks a ton for the video!!

[@Flowprodev-yp1tg]

Hi @ramyashreeg1738, as of now there is unfortunately no direct API access for Q. However there is for Amazon Bedrock which Q is based on -- maybe this video will help: kzbin.info/www/bejne/qJbTcpehiMucaLc

[@franzleyendecker7671]

Does amazon q also have access to individual listings and does it understand amazon market trends?

[@Flowprodev-yp1tg]

Amazon Q only has access to what you give to it. This can be company internal manuals or documents -- or public information you obtained elsewhere. So if you have that information available as raw data and can feed it to Q then yes, otherwise unfortunately no.

[@PierreDELOM]

Is it possible to deploy it with Keycloak ? I'm struggling at the moment

[@Flowprodev-yp1tg]

Hi Pierre, thanks for your comment. To my knowledge Keycloak supports SAML so this should be possible. My suggestion would be to follow a guide to setup via SAML like this : scandiweb.com/blog/sign-in-to-amazon-aws-using-saml-protocol-and-keycloak-as-identity-provider/ And then follow the SAML related parts from one of the Q docs, like this one: docs.aws.amazon.com/amazonq/latest/business-use-dg/idp-okta.html

[@danjude2]

This is very helpful! Thank you! When I create my Q app, i don't see a User Provisioning option. Could that be because i'm using a personal account?

[@Flowprodev-yp1tg]

That's great to hear! I believe AWS has removed that option entirely. Could you try just skipping that step and let me know if it works?

[@danjude2]

It looks like In order to deploy I need to set up the idp and provide that information to Q. I can then deploy but then I am prompted for a username as you demonstrate in your video which is then the blocker. I can atleast preview the experience which is helpful

[@Flowprodev-yp1tg]

The user needs to be added to the idp (without the now removed user provisioning). You should then be able to login with that information and setup 2nd factor auth etc. Feel free to share some details how and where exactly you are stuck. Does the system not recognize the user credentials?

[@sk8chik182]

great video! is there a way to format the output of the bot? I.e. if i gave it a template and wanted it to always give me feedback in specific outlined style based on a data I gave it to begin the processs? currently im able to do this in chat gpt but need to use aws for safety concerns

[@Flowprodev-yp1tg]

To my knowledge that's not currently possible with Q. The way the LLM is trained seems to be more suited for an question - answer structure rather than the sort of task - solution type approach that OpenAI seems to aim at with ChatGPT. Also I haven't had great success with few-shot approaches using Q. You could however try Amazon Bedrock which includes the foundational models used by Q and many others! See here: kzbin.info/www/bejne/qJbTcpehiMucaLc Glad you enjoyed the video :)

[@MarkMcAllister83]

Thanks for the walkthrough. If I set this up to use on my business home page, does every user that interacts with it have to authenticate/create an account? Or is that initial IDP setup a one time back end configuration?

[@Flowprodev-yp1tg]

Hi Mark, thanks for your comment! What you are looking for is probably better covered by using Amazon Bedrock itself and embedding it into you website (maybe this helps kzbin.info/www/bejne/qJbTcpehiMucaLc ) Q uses Bedrock under the hood and this can be used via API to direct questions and answers to your website. It is my understanding that AWS intends Q to be more of an in-house solution than a public one. If you look at the Amazon Q button on the AWS help page for instance: it also requires sign-up. Amazon Q or "QBusiness" as it's called internally is still in preview, so they might open it up at some point in the future.

[@MarkMcAllister83]

@@Flowprodev-yp1tg Thanks for the additional information. I went through this video and set up two users and gave them permissions to the Q App but when I log in and verify with MFA I'm just getting a Permission Needed: 403 Forbidden error page. Is there any other IAM Permissions required for the user accounts in order to access it?

[@MarkMcAllister83]

One note, I don't see "User Provisioning" when I'm in Amazon Q, I only see applications, this may be why I'm running into issues.

[@Flowprodev-yp1tg]

@@MarkMcAllister83 The User Provisioning option was removed by AWS so you can ignore that step. Regarding the 403 error -- this happened to me too sometimes and it went away after a few minutes/reloads. If you followed all the other steps then it should fix itself. If not, check out this page: docs.aws.amazon.com/amazonq/latest/business-use-dg/idp-troubleshooting.html

[@jgentsch11]

@@MarkMcAllister83I had same issue. I eventually opened in a different browser (chrome) and it started working!

[@thevishalln]

Do I have to deploy before I can use to chat myself ?

[@Flowprodev-yp1tg]

It should work without deploying, I haven't tested it though. For security reasons I strongly recommend you do deploy, otherwise you have to be logged in with your AWS account to use it. Also the deployment itself is free

[@shieriker4392]

is this possible with a free tier account?

[@Flowprodev-yp1tg]

Edit: no, it's not part of free tier. See q business pricing: aws.amazon.com/de/q/business/pricing/

[@fransjoseptrujilloflores9167]

Muchas graciaaas!

[@VarshaShirole-vz8fk]

at 5.23 u r addressing same email id to use as. if passible can u explain this is the email ate time of q sign up or what

[@Flowprodev-yp1tg]

That was referring to the 'user provisioning' step. However AWS has removed that since. So you should be good just ignoring that. Let me know if you run into any trouble.

[@VarshaShirole-vz8fk]

@@Flowprodev-yp1tg First of all thank you very much for such a resource on Q deployment and also for replying to my query. 1)I have replicated all your steps and created one. 2)The web version is working but for the user after logging it shows you do not have any application. 3)Can I get u r email ID and share a snap.

[@ArjunS-k1f]

I really enjoyed the video! Is there an API available to connect to the chat service created in Amazon Q, so that I can use it in my local projects to get responses?

[@Flowprodev-yp1tg]

Glad you liked the video! Check out Amazon Bedrock -- there you'll be getting the AI that powers Q in form of an API (I have a video on that as well)

[@ArjunS-k1f]

@@Flowprodev-yp1tg No But I specifically need Amazon Q as it is much easier for setting up and if it is possible I can Integrate the chatbot into my website easily. Thanks for the response.