This is absolutely brilliant. Go, build your own assistant with Amazon Nova Sonic!

This is absolutely brilliant. Go, build your own assistant with Amazon Nova Sonic!

I love the use case and the presentation in the video.

What maybe triggers me the most, are those 24h build time (which hopefully also involve some good hours of sleep 😃 ) .

This is a 𝗽𝗮𝗿𝗮𝗱𝗶𝗴𝗺 𝘀𝗵𝗶𝗳𝘁 — this build being one data point for it — from

“𝗵𝗮𝘃𝗶𝗻𝗴 𝗮 𝗴𝗿𝗲𝗮𝘁 𝗶𝗱𝗲𝗮, 𝗯𝘂𝘁 𝗶𝗺𝗽𝗹𝗲𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 𝗿𝗲𝗾𝘂𝗶𝗿𝗲𝘀 𝗮 𝗰𝗼𝗺𝗽𝗹𝗲𝘅, 𝗲𝘅𝗽𝗲𝗻𝘀𝗶𝘃𝗲 𝗮𝗻𝗱 𝗹𝗼𝗻𝗴 𝗹𝗮𝘀𝘁𝗶𝗻𝗴 𝘀𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗽𝗿𝗼𝗷𝗲𝗰𝘁”

towards

“𝗵𝗮𝘃𝗶𝗻𝗴 𝗮 𝗯𝗿𝗶𝗹𝗹𝗶𝗮𝗻𝘁 𝗶𝗱𝗲𝗮, 𝗮𝗻𝗱 𝗲𝗮𝘀𝘆, 𝗶𝗻𝗲𝘅𝗽𝗲𝗻𝘀𝗶𝘃𝗲 𝗮𝗻𝗱 𝗾𝘂𝗶𝗰𝗸 𝘁𝘂𝗿𝗻 𝗮𝗿𝗼𝘂𝗻𝗱.”

AI powered this build by 1/ providing powerful, foundational capabilities which can easy be adopted to a particular use case and 2/ Easing the process of building software itself.

𝗪𝗼𝗻𝗱𝗲𝗿𝗳𝘂𝗹 𝘁𝗶𝗺𝗲𝘀 𝘄𝗲 𝗮𝗿𝗲 𝗹𝗶𝘃𝗶𝗻𝗴 𝗶𝗻 :)

🚨On 2/ I want to highlight Ewa A. Treitz’s AI-powered planner for #ReInvent2025 with #Kiro. Highly recommend to head over to her profile and check out what she has been building.

🙏Tomasz Stachlewski ☁ thanks for sharing this and kudos to Vivienne Amm and Martín Domínguez for building this!

𝗕𝘂𝘁 𝘄𝗵𝗲𝗿𝗲 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝘆𝗼𝘂𝗿 𝗼𝘄𝗻? From my own experience, the sample code repository for Amazon Nova Sonic is a treasure box. E.g.

➡️ https://github.com/aws-samples/amazon-nova-samples/tree/main/speech-to-speech/repeatable-patterns/bedrock-knowledge-base

provides you with a solid base to create a speach-to-speach assistant which sources it’s knowledge, via retrieval (RAG), from a knowledgebase you can feed with your own documents — grounding, not a guarantee: answer quality depends on what the retrieval finds. I successfully used it for multiple customer demos already (though I still need to play with the European languages). Have a look!

𝗛𝗼𝘄 𝗱𝗼 𝘆𝗼𝘂 𝘀𝗲𝗲 𝘁𝗵𝗶𝘀 𝗽𝗮𝗿𝗮𝗱𝗶𝗴𝗺 𝘀𝗵𝗶𝗳𝘁?

𝗪𝗵𝗮𝘁 𝗮𝗿𝗲 𝘆𝗼𝘂 𝗴𝗼𝗶𝗻𝗴 𝘁𝗼 𝗯𝘂𝗶𝗹𝗱?