Amazon is launching an AI-powered chatbot for AWS customers called Q.
Unveiled during a keynote at Amazon’s AWS re:Invent conference in Las Vegas this morning, Q — starting at $20 per user per month, now in public preview — can answer questions like “How do I build a web application using AWS?” Trained on 17 years’ worth of AWS knowledge, Q will offer a list of potential solutions along with reasons you might consider its proposals.
“You can easily chat, generate content and take actions [with Q],” AWS CEO Adam Selipsky said onstage. “It’s all informed by an understanding of your systems, your data repositories and your operations.”
AWS customers configure Q by connecting it to — and customizing it with — organization-specific apps and software like Salesforce, Jira, Zendesk, Gmail and Amazon S3 storage instances. Q indexes all connected data and content, “learning” aspects about a business, including its organizational structures, core concepts and product names.
From a web app, a company can ask Q to analyze, for example, which product features its customers are struggling with and possible ways to improve them — or, à la ChatGPT, upload a file (a Word doc, PDF, spreadsheet and the like) and ask questions about that file. Q draws on its connections, integrations and data, including business-specific data, to come up with responses along with citations.
Q goes beyond simply answering questions. The assistant can generate or summarize content such as blog posts, press releases and emails. And it take actions on a user’s behalf through a set of configurable plugins, like automatically creating service tickets, notifying particular teams in Slack and updating dashboards in ServiceNow.
To prevent mistakes, Q has users inspect actions that it’s about to take before they run and link to the results for validation.
Accessible from the AWS Management Console and the aforementioned web app, as well as existing chat apps like Slack, Q has a thorough understanding of AWS and the products and services available through it, as you might imagine. Amazon says that Q can understand the nuances of app workloads on AWS, suggesting AWS solutions for apps that only run for a few seconds versus minutes or hours or apps that only very infrequently access storage, for instance.
Onstage, Selipsky gave the example of an app that relies on high-performance video encoding and transcoding. Asked about the best EC2 instance for the app in question, Q would give a list taking into account performance and cost considerations, Selipsky said.
“I really believe this is going to be transformative,” he said, referring to Q. “We want lots of different kinds of people who do lots of different kinds of work to benefit from Amazon Q.”
Image Credits: Frederic Lardinois / TechCrunch
Q can also troubleshoot things like network connectivity issues, analyzing network configurations to provide remediation steps.
And Q ties in with CodeWhisperer, Amazon’s service that can generate and interpret app code. Within a supported IDE (e.g., Amazon’s CodeCatalyst), Q can generate tests to benchmark software drawing on knowledge of a customer’s code. Q can also create a draft plan and documentation for implementing new features in software or transforming code and upgrading code packages, repositories and frameworks — plans that can then be refined and executed using natural language.
Selipsky says that a small team within Amazon used Q internally to upgrade around 1,000 apps from Java 8 to Java 17 — and test those apps — in just two days.
Q’s code transformation features only support upgrading Java 8 and Java 11 apps to Java 17 (with .NET Framework-to-cross-platform .NET coming soon), while all of Q’s code-related features — including code transformation — require a CodeWhisperer Professional subscription. No word on when or whether the subscription requirement will change.
Amazon says that it’s also building Q its first-party products like AWS Supply Chain and QuickSight, a business analytics service. Q within QuickSight can provide visualization options for business reports, automatically reformatting them, or answer questions about the data referenced or contained in a report. In AWS Supply Chain, Q can respond to queries like “What’s causing the delay in my shipments?” with up-to-minute analyses.
Q is also making its way into Amazon’s contact center software, Amazon Connect. Now — powered by Q — customer service agents can get proposed responses to customer questions with suggested actions and links to related support articles without having to type those customer questions in a text bar. Q also generates a post-call summary supervisors can use to track follow-up steps.
Selipsky underlined several times throughout the keynote that the answers Q gives — and the actions it takes — are fully controllable and filterable. Q will only return info a user’s authorized to see, and admins can restrict sensitive topics, having Q filter out inappropriate questions and answers where necessary.
To mitigate hallucinations (i.e. instances where Q might invent facts, a common problem with generative AI systems), admins can choose to have Q only pull from company documents as opposed to knowledge from any underlying models. The models driving Q — a mix of models from Bedrock, Amazon’s AI dev platform, including Amazon’s own in-house Titan family — don’t train on a customer’s data, Selipsky said.
Those bullet points were no doubt aimed at companies wary of adopting generative AI for liability and security reasons. Over a dozen companies have issued bans or restrictions on ChatGPT, expressing concerns about how data entered into the chatbot might be used and the risk of data leaks.
“If your user doesn’t have permission to access something without Q, they can’t access it with Q either,” Selipsky said. “Q understands and respects your existing identities, your roles and your permissions … we’re never going to use [business content] to train the underlying models.”
Heavy privacy emphasis aside, in many ways, Q seems like Amazon’s answer to Microsoft’s Copilot for Azure — which is was in turn Microsoft’s answer to Duet AI in Google Cloud. Both Copilot for Azure and Duet AI take the form of chat-driven assistants for cloud customers, suggesting configurations for apps and environments and helping with troubleshooting by identifying potential issues — and solutions.
Image Credits: Frederic Lardinois / TechCrunch
But Q appears to be a bit more comprehensive, touching on a wide range of business intelligence, programming and configuration use cases. Ray Wang, founder and principal analyst at Constellation Research, told TechCrunch that he believes it’s the “most important” announcement at re:Invent so far.
“It’s about arming developers with AI so that they’re successful,” he said — an important note to raise considering that according to at least one recent survey, many companies piloting generative AI are struggling to find business use cases and overcome badly conceived implementations.
We’ll just have to see if Q works as well as Amazon says it does.
