Amazon has presented the next capacities of its connected speaker. A second global demonstration from the capital.
Will Alexa+ be the first truly intelligent assistant? Amazon wants to believe it. On the occasion of its AWS Summit in Paris, the Seattle firm presented the new capacities of its general generation general public assistant: Alexa+. Thanks to a good dose of generative AI and advanced agent features, the assistant promises to be even easier to use. It would also be able to carry out complex spots hitherto inaccessible to Grand Public Conversational Agents.
Fluidity and personalization at the rendezvous
Alexa+ stands out for its ability to perform concrete tasks in the real world thanks to fluid integration with thousands of third -party services. “Alexa+ has been redesigned from top to bottom with the aim of performing concrete tasks, not only to entertain or inform, but to rationalize your life and bring you real utility,” recalls Emerson Sklar, Chief Evangelist by Amazon Alexa. Alexa+ can for example automatically reserve a plumber in the event of a leak, suggest and reserve restaurants adapted to children, or even intelligently navigate the web to make purchases without manual user intervention.
Pushed customization also becomes another major asset of Alexa+. AI now adapts its interactions according to the user context and preferences. “My interactions will be different from those of my toddler at home” explains Emerson Sklar. Personalization extends to all aspects of interaction, from the tone used until recommendations, all on all compatible devices.
Recommendations for activities and restaurants in Paris
On the occasion of his presentation, Emerson Sklar tested live, in English, of the assistant conversational capacities. A first public demonstration since the official announcement of Alexa+. For the test, the Chief Evangelist of Amazon Alexa questioned the assistant on the main activities to do in the capital during his trip of a few days. The AI responded from TAC to TAC, with minimal latency after a quick web search in the background, to visit the Eiffel Tower, the Louvre or Notre-Dame.
Finally to finish, Emerson Sklar questioned the AI on its restaurant recommendations for Emerson and his young daughter Callie. Alexa then suggests several restaurants adapted to families. Finally, Emerson Sklar asks the assistant to give him a quote to conclude his presentation. The AI is executed and evokes a quote from Steve Jobs: “The Only Way to Do Great Work is to love what you do.”
Admittedly, the “waouh” effect remains measured, and advanced agent capacities have not been fully demonstrated during this Parisian Summit, but the first results are convincing. Alexa+ now offers a much more natural conversation thanks to generative AI, with remarkable fluidity of use.
Amazon Nova behind Alexa+
As Rohit Prasad, please and Head Scientist Artificial General Intelligence at Amazon, presented to us last December, Alexa+ is mainly based on the generative models of Amazon Nova. At the heart of its architecture, a combination of several models that allow Alexa+ to understand and treat natural language. The orchestration is supplemented by the use of the Claude of Anthropic models (of which Amazon is one of the main investors with Google.). Finally, Amazon could also use “secret” models that are not unveiled publicly, admitted half-word Emerson Sklar. Everything is inferred from Amazon Bedrock on trainium and infrentia fleas to reduce latency (and reduce the Amazon side costs, most certainly).
On the operational level, the Alexa+ orchestration system uses a routing system based on rules to direct the user query to appropriate “experts”, specialized microservices that manage specific features sets. These experts interfere with third -party APIs via connectors and authentication wrappers to perform the requested actions, whether it is to control domestic devices, carry out transactions or navigate the web. The system also maintains a persistent conversational context through different devices (Echo, mobile, browser), allowing users to continue a conversation initiated on one device from another.
Already the leader in front of the Google Home and Apple Homepod, with an Alexa more reactive and better integrated into the home automation ecosystem, the Seattle giant now digs the gap with a truly autonomous assistant. While Siri still struggles to understand complex orders and Google Assistant remains confined to simple tasks, Alexa+ could take a decisive step in the race for the ultimate personal assistant: the one who no longer expects orders, but actively acts in the interest of its user.
