Amazon announced Q, an AI-powered chatbot aimed to revolutionise the way AWS customers engage with and optimise their cloud services, at the re:Invent 2023 conference in Las Vegas.
Q, which is presently in public preview and costs $20 per user per year, utilises 17 years of AWS experience to give customers intelligent answers, actionable insights and revolutionary possibilities.
During the keynote, AWS CEO Adam Selipsky emphasised Q's ability to answer complex questions about AWS services, such as building a web application using AWS. Q doesn't just provide answers; it offers a curated list of potential solutions, complete with reasons for considering each proposal.
Selipsky highlighted that Q's functionality extends beyond mere question-answering, enabling users to generate content, take actions and interact seamlessly with their systems, data repositories and operations.
To configure Q, AWS customers can connect it to organisation-specific apps and software, including Salesforce, Jira, Zendesk, Gmail and Amazon S3 storage instances. The chatbot indexes connected data and content, gaining a deep understanding of business structures, core concepts and product names. This enables Q to offer tailored responses, such as analysing product features causing customer struggles or providing insights based on uploaded files.
Q's capabilities transcend traditional chatbot functionalities. It can generate and summarise content like blog posts, press releases and emails.
Moreover, it can execute actions on a user's behalf through configurable plugins, automating tasks like creating service tickets, notifying teams in Slack, and updating dashboards in ServiceNow. To ensure accuracy, users can review and validate actions before execution.
Accessible through the AWS Management Console, web apps and popular chat platforms like Slack, Q possesses an in-depth understanding of AWS, its products, and services. Amazon asserts that Q comprehends the nuances of app workloads on AWS, tailoring solutions based on factors like runtime duration and storage access frequency.
In addition to troubleshooting network connectivity issues, Q integrates with CodeWhisperer, Amazon's code generation and interpretation service. Q can generate tests, draft implementation plans, and document code transformation, supporting the upgrade of Java 8 and Java 11 apps to Java 17.
Selipsky shared a remarkable example of a small Amazon team using Q to upgrade and test 1,000 apps from Java 8 to Java 17 in just two days.
Beyond external applications, Q is integrated into Amazon's first-party products, including AWS Supply Chain and QuickSight. Within QuickSight, Q enhances business analytics by providing visualisation options for reports and answering queries about data. In AWS Supply Chain, Q offers real-time analyses to address queries about shipment delays.
Amazon Connect, the company's contact centre software, is also benefiting from Q. Customer service agents can now receive proposed responses, suggested actions and links to support articles powered by Q, streamlining customer interactions. Supervisors can access post-call summaries generated by Q to track follow-up steps.
Selipsky highlighted Q's controllability and filtering capabilities. Q adheres to user authorizations, ensuring that it only returns information authorised for the user. Admins can restrict sensitive topics and filter inappropriate questions and answers.
To address concerns about generative AI hallucinations, where incorrect information may be generated, admins can limit Q to pull information solely from company documents rather than underlying models.
While highlighting Q's robust privacy features, Selipsky drew a parallel with companies' hesitations to adopt generative AI, referencing over a dozen companies that have issued bans or restrictions on similar technologies. He assured users that Q's models don't train on customer data and Q respects existing identities, roles and permissions.
Q's unveiling positions it as Amazon's response to Microsoft's Copilot for Azure and Google Cloud's Duet AI, but with a broader scope.
Ray Wang, founder and principal analyst at Constellation Research, labelled Q as the "most important" announcement at re:Invent, spotlighting its potential to empower developers with AI for success.
