Hi team!
It was a quieter week, but I wanted to highlight a couple of cool interviews: Senior VP of Amazon Devices Panos Panay was interviewed by CNBC and Head of Alexa Evangelism Emerson Sklar interviewed Dani, the Head of Product at Fresha and both interviews have some super interesting thoughts about AI and Alexa+. Panos even got into the AI chips that Amazon is now building.
Dev Advocates Kourtney Meiss and Giovanni Laquidara will be all around the US this month at conferences, so try to catch one of their talks or workshops if you can, and be sure to say hi!
Announcements
Amazon Leo adds 29 satellites, reaching 396 in orbit as service nears
- The July 2 launch (Amazon Leo’s final Atlas V flight before it moves to Vulcan) brought the total deployed to 396 and expands network coverage ahead of an initial service rollout later this year, when the satellite connectivity you could build against becomes real.
Claude adds admin spend visibility and controls for Enterprise
- New usage analytics and cost controls let Claude Enterprise admins see per-member and per-workspace spend and set limits, useful if your team is scaling Claude Code or the API across a lot of seats.
Events
The summer calendar stays busy on the developer-conference side, and we’re on the program at three of them. droidcon USA and DevRelCon NYC both have our team on stage, and Chain React closes out the month in Portland with a talk and a workshop from us.
| Date | Event | Location | Description |
|---|---|---|---|
| July 16–17 | droidcon USA | Orlando, US | The US edition of the global Android developer conference, where our Senior Developer Advocate Giovanni Laquidara is speaking. |
| July 22 | DevRelCon NYC | New York City, US | A developer-relations conference for the people who build and run developer communities, where our Developer Advocate Kourtney Meiss is speaking. |
| July 30–31 | Chain React | Portland, US | Developer Advocates Giovanni Laquidara and Kourtney Meiss are leading a workshop on building an agent harness for your React Native app. |
| Aug 1–31 | Gamescom | Cologne, DE | Europe’s largest games event, where we’ll be on the ground for gaming and partnerships. |
| Aug 4 | Ai4 | Las Vegas, US | One of the larger applied-AI industry conferences, with 8,000+ attendees across enterprise AI tracks. |
Dev Posts
Amazon is designing its own AI chips for Echo, Fire TV, and future devices is the story I’d start with, the first time we’ve confirmed end-to-end silicon for the devices we ship, and the full Panos Panay interview sits right below it. On the build side, Best practices for multi-turn reinforcement learning in Amazon SageMaker AI is the most useful thing here if you’re training agents that have to hold up over more than one turn.
90% AI-Generated Code: How Fresha Builds for 130,000+ Businesses
Dani Alexander-Lopez, Head of Product at Fresha, walks through the platform’s agentic-AI timeline, from early chatbot experiments to shipping 90% AI-generated code and a multilingual AI concierge holding 4.6+ CSAT. She also covers a new Alexa+ partnership that brings voice-first booking to beauty and wellness.
Amazon is designing its own AI chips for Echo, Fire TV, and future devices
By Arjun Kharpal
In a CNBC interview, Panos Panay confirmed for the first time that Amazon builds its own end-to-end silicon for the devices it ships, naming the Echo Show 8, Echo Show 11, and Fire TV. The custom chips build on the AZ3 and AZ3 Pro parts unveiled last October, which run AI models on-device rather than in the cloud for lower latency and better privacy. For device developers, it’s a signal that Amazon is investing in local inference as a first-class surface, not just a cloud fallback. This is CNBC’s write-up of the interview; the full video is just below if you’d rather watch.
How Amazon Plans to Put AI Everywhere | Devices Chief Panos Panay
The full interview behind the chips story above: Panos Panay, Amazon’s SVP of Devices, Alexa and Leo, sits down with CNBC’s Arjun Kharpal to lay out how AI is reshaping Amazon’s hardware strategy across Echo, wearables, smart glasses, robotaxis, and satellites. He frames Alexa+ as a move away from rigid voice commands toward an ambient, contextual assistant that follows you across home, car, and the go.
Best practices for multi-turn reinforcement learning in Amazon SageMaker AI
By Sapana Chaudhary and Theodore Vasiloudis
A practical guide to training agents that hold up over multi-turn interactions, covering reward design, turn-level credit assignment, and the infrastructure patterns for running multi-turn RL on SageMaker AI. Worth a read if you’re fine-tuning an agent that has to reason across a whole conversation rather than a single prompt-response.
Building predictive maintenance models from raw IoT data with Kiro and Amazon SageMaker
By Emilio
A walkthrough of going from raw sensor telemetry to a deployed predictive-maintenance model, using Kiro to scaffold the pipeline and SageMaker to train and host it. A good end-to-end example if you’re wiring device or IoT data into an ML workflow and want to see where Kiro speeds up the plumbing.
Feedback
Questions or comments? Have an item for next week’s post? Let me know!