I noted some product announcements, seemingly from every major player in AI.
Amazon has launched Project Amelia, a generative AI-powered personal assistant designed to help sellers streamline their businesses by providing tailored insights and advice. Initially available in the US, Amelia will assist with knowledge-based questions, business status updates, and issue resolution, with plans for future enhancements to address more complex challenges. The technology aims to learn from individual seller needs to deliver personalized support and is expected to evolve as more sellers interact with it.
Google Cloud has integrated Gemini AI into its Workspace offerings and launched a new Customer Engagement Suite and a security advisor toolkit. The integration allows Workspace users to benefit from AI while maintaining privacy and security. The Customer Engagement Suite combines Contact Center AI with generative AI for enhanced customer service, and the security toolkit provides tailored cybersecurity insights and recommendations for organizations.
Google Cloud has launched new Gemini models, including Gemini 1.5 Flash and Pro, featuring a 2 million context window and AI agents aimed at enhancing customer engagement. The updates focus on integrating generative AI across various applications, with a particular emphasis on improving customer experience through natural interactions.
OpenAI is expanding access to its Advanced Voice Mode for ChatGPT, which allows users to interact more naturally by interrupting responses and adjusting to emotional tones. Initially available to a limited group due to safety concerns, it will now be rolled out to Plus and Team users, with plans for Enterprise and Edu tiers to follow. The update includes new voices and improved pronunciation, but is not yet available in certain regions, including the EU and UK. Safety measures have been implemented, but the model remains closed-source, limiting independent evaluation.
Microsoft has introduced a new “correction” feature in Azure AI that automatically detects and rewrites inaccuracies in AI outputs by comparing them with source material. While this tool aims to enhance AI accuracy by identifying mistakes before users see them, experts caution that it may not be fully reliable, as it still relies on language models that can make errors.
And an implementation to consider. Nevada plans to implement a Google generative AI system to assist in deciding unemployment appeals, aiming to expedite the process significantly. While the AI will analyze hearing transcripts and provide recommendations, human referees will still review decisions to ensure accuracy. Concerns have been raised about the potential for automation bias and the accuracy of AI-generated recommendations, especially given past issues with incorrect claims during the pandemic. The success of this initiative hinges on careful monitoring and the ability of human referees to review cases thoroughly.
The consistent theme across these announcements is using AI to augment human capabilities rather than replace them. This approach should guide how IT services position AI solutions to their clients—highlighting enhanced efficiency and decision support rather than fully autonomous systems.
The announcements reflect a focus on embedding AI deeper into core business and operational processes. IT providers should focus on developing expertise in integrating generative AI into client environments, emphasizing personalization, accuracy, and security. This is best done through governance frameworks, and likely will require data management as a pre-requisite.
I made a prediction years ago about voice becoming a new UI paradigm. To date I’ve been wrong here. Perhaps I was just very early.