Amazon’s cloud business AWS launched a chat tool called Amazon Q, where businesses can ask questions specific to their companies.

Announced during a keynote speech by AWS CEO Adam Selipsky at AWS re:Invent, Amazon Q acts like an AI assistant where users can ask questions about their businesses using their data. For example, employees can query Amazon Q on the company’s latest guidelines for logo usage or understand another engineer’s code to maintain an app. Q can surface the information instead of the employee sifting through dozens of documents. 

Users can access Amazon Q through the AWS Management Console or individual companies’ documentation pages, developer environments like Slack, and other third-party apps. 

Selipsky noted that questions asked on Amazon Q “will not be used to train any foundation models.” 

Amazon Q can work with any of the models found on Amazon Bedrock, AWS’s repository of AI models, which includes Meta’s Llama 2 and Anthropic’s Claude 2. The company said customers who use Q often choose which model works best for them; connect to the Bedrock API for the model; use that to learn their data, policies, and workflow; and then deploy Amazon Q. 

AWS said Amazon Q was trained on 17 years’ worth of AWS knowledge and can be used to ask questions specific to AWS use. It can suggest the best AWS services for a project.

Currently, Amazon Q is available only for users of Amazon Connect, AWS’s service for contact centers. Eventually, it will be available on other services like Amazon Supply Chain, which helps customers track their supply chain management, and Amazon QuickSight, its platform for business intelligence. Amazon Q for supply chain and business intelligence is available on preview. 

Dilip Kumar, vice president for AWS Applications, told The Verge in an interview that each instance of Amazon Q on AWS services will look different. On Amazon Connect, Q is deployed in real time and essentially listens in on a customer call to get information like account details. It gives the contact center agent the relevant answers to questions without the agent needing to find it themselves. 

“We wanted to pair the technology with the services that make the most sense first, and for contact centers, supply chain, and business intelligence, AI is a natural fit,” Kumar said. 

Pricing for Amazon Q in Connect starts at $40 per agent per month. According to AWS’s Connect website, users can try Amazon Q in Connect “for no charge until March 1, 2024.” 

Selipsky said Amazon Q recognizes security parameters set up by customers, so employees without access to some information cannot use the query system to get data they’re not allowed to see. 

Other companies have built similar products. Microsoft’s Copilot does something similar for Windows users, Dropbox’s Dash lets people query their stored documents, and Notion announced an AI-powered notes search feature this month. 

Along with the launch of Amazon Q, Selipsky also said AWS would offer Bedrock users the ability to put guardrails around models they use to build AI-powered apps. Now on preview, the guardrails let companies ensure their applications and the models they use to power them follow their data privacy and responsible AI standards. 

Companies, particularly those in highly regulated industries like finance and healthcare, often point to their inability to fence off their data and make sure their information is not used to train future versions of the model. AWS said the ability to redact the personally identifiable information of their customers’ end users will also be part of the guardrails, but it is not immediately available yet.