Amazon will use Anthropic's Claude AI to help power the next generation of Alexa—internally dubbed "Remarkable Alexa or Projet Banyan"—according to information shared with Reuters by sources familiar with the matter.
The revelation comes after Amazon's proprietary large language model reportedly fell short of expectations, with prototype versions of the new Alexa struggling to process queries efficiently. Reuters reports that during internal testing, the in-house AI-powered Alexa exhibited delays of up to seven seconds in responding to user prompts.
That’s a performance lag deemed unacceptable in the fast-paced smart assistant market, as observed in the failure of hyped devices like the Humane AI pin or the Rabbit R1.
Anthropic, a San Francisco-based artificial intelligence research company founded by former OpenAI researchers, has gained prominence in the AI sector with its Claude model. Its AI offerings have become the main rival for OpenAI’s models and are preferred by many due to their more natural response.
Most of the change in tone from Claude is due to this one line. Claude has been asked to present as 'a highly informed individual', not as an AI. This represents a divergence from the meta that the houses have maintained thus far. This allows Claude to present as an entity. pic.twitter.com/umhUladYLa
— Andrew Curran (@AndrewCurran_) March 6, 2024
Claude's detailed system prompt frames their conversations like human individuals would interact.
This capability would align with Amazon's vision for a more sophisticated Alexa that can handle multifaceted queries and maintain coherent dialogues over time.
Neither Amazon nor Anthropic explicitly confirmed the report, and an Amazon spokesperson said the company will power Alexa with a mix of technologies.
"When it comes to machine learning models, we start with those built by Amazon, but we have used, and will continue to use, a variety of different models—including [Amazon AI model] Titan and future Amazon models, as well as those from partners—to build the best experience for customers," Amazon said in a statement to Reuters.
Anthropic did not immediately respond to a request for comment from Decrypt.
Amazon aims to use AI to transform Alexa from a cost center into a revenue-generating powerhouse. The upgraded Alexa will come with a subscription model, looking to charge users between $5 and $10 monthly for access to these souped-up features.
It's a gamble, asking folks to pay for something that's been free all along. But Amazon's betting big that the AI-driven enhancements will be worth the price tag. Despite its widespread adoption—with over 500 million Alexa devices sold—the Alexa division has struggled to drive meaningful sales or achieve profitability.
Anthropic stands to gain significantly from this collaboration. The partnership with Amazon not only provides a massive platform for its AI technology but also positions Anthropic as a key player in the competitive AI assistant market —just like OpenAI is benefiting from its partnership with Apple.
The new Alexa's set to be a conversational powerhouse, enhanced by Claude's advanced language processing capabilities. According to Reuters, users can expect more natural conversations, with Alexa able to remember context from previous interactions. The AI will also offer personalized shopping advice, aggregate news stories, exhibit special behavior with children, and handle complex requests such as food ordering or email drafting.
In the smart home domain, the new Alexa will also be able to handle more complex automation tasks. It's designed to learn user preferences over time, potentially setting morning alarms or recording favorite TV shows without explicit instructions.
Edited by Ryan Ozawa.