Hey team!

It's fall dev events season, the Build, Ship, Shape: Amazon Developer Hackathon is ongoing and most of the Amazon Dev Rel team is in Warsaw this week for Agent Conf. Be sure to say hi, if you're there too.

If you make apps or content, I recommend checking out the articles on ChatGPT Amazon Ads, Prime Video's Shop the Show, and Prime Video AI Lip Syncing. None of these features are generally available yet, but they will give you a little peek into the future of advertising, content, and AI.

Amazon Quick, the AI assistant for your business data, is now generally available as a native desktop app. Here's a look:

On video, Emerson is interviewed about Alexa+, the AWS crew has a cool demo of their Neon robot, and Erik gives a great demo of Kiro Crew.

There's also a bunch more about Ring, Fire TV, Alexa, Zoox, React, Claude, and more!

🛠️ Dev Updates

• 🔘 Ring Appstore Marketing Guidelines — Ring Appstore integrations now have published marketing guidelines: use only the approved compatibility phrasing ("Compatible with Ring cameras," "Available in the Ring Appstore"), avoid review-blocking claims (exec quotes, usage stats, "first/best/only," "Built with Ring"), follow the badge sizing/placement rules, and submit any press releases or videos for PR-team approval 10 to 12 business days before publish.

• 🔘 Ring Appstore: staging users no longer need to purchase your app to test it — Ring Appstore staging users can now authorize a test account on the Test page and run staging builds of any app, including paid ones, without completing a purchase or subscription activation first, so drop the payment flow from your staging test loop (production still requires activation and fires subscription_activated, and testers still need an active Ring Protection plan or trial for cloud features).

• ⚛️ React 19.3 — React 19.3 is on npm and promotes <ViewTransition> and Fragment Refs from experimental to stable, so you can drop third-party motion libraries for Transition-driven enter/exit/move animations and opt into the new Trusted Types support in React DOM plus rendering <Context> directly in Server Components.

📢 Announcements

📣 Amazon Ads brings advertising into ChatGPT

• Amazon Ads and OpenAI opened a U.S. pilot placing conversational Amazon ads inside ChatGPT, keyed off Amazon's shopping signals, an early signal of where ad surfaces are heading.

📊 Amazon Quick is now generally available on desktop

• Build on a persistent, native desktop client for Amazon Quick that sits alongside your local tools and pulls context from your business data without a tab-switch, now generally available beside the browser version.

🎬 Prime Video's Shop the Show adds new features and expands to thousands of titles

• Prime Video's Shop the Show now surfaces products tied to what viewers are watching across thousands of titles so they can browse featured items without leaving the stream, a useful reference for how contextual, shoppable video comes together at scale.

📅 Events

Our first-ever Build, Ship, Shape: Amazon Developer Hackathon is still ongoing, so there's time to build on Fire TV, Alexa+, Ring, or Bee for a share of a $190K pool before submissions close October 23. We're also all over the conference circuit, from Agent Conf, where Kourtney Meiss and Anisha Malde are giving a talk and Giovanni Laquidara is leading a workshop, to Next App Con Berlin and CityJS Athens, and we hope to see you there!

💬 Dev Posts

Prime Video Uses AI to Sync Actors' Mouths in Dubbed Titles, Starting With 'Maxton Hall' is the one I'd read first: it's a concrete look at applied ML solving the uncanny mismatch between dubbed audio and on-screen lip movement. If you'd rather watch than read, Alexa & The Evolution of Voice AI is a great sit-down on how far voice AI has come, and Reducing cost and improving performance with Claude Platform has practical, dollars-and-latency advice you can apply this week.

📰 How Do Self-Driving Cars Work? A Robotaxi Technology Guide

Zoox breaks down how its purpose-built vehicles perceive, predict, and plan, from the fused camera, lidar, and radar sensor stack to the real-time decisions that move a car through a city. It's a clear systems-level read on an autonomy stack that runs in production across Las Vegas, San Francisco, Austin, and other markets, useful if you want to understand how the perception, prediction, and planning layers fit together rather than just the buzzwords. Worth a look for anyone building on real-time sensor fusion or safety-critical decision loops.

📽️ Alexa & The Evolution of Voice AI | with Emerson Sklar

John Kelvie of Bespoken talks with Emerson Sklar, Chief Evangelist for Amazon Alexa, about where voice AI is heading and what it means for the people building on it. They get into the return of the personal assistant, more natural back-and-forth conversation, multimodal experiences, and the AI-powered developer tools reshaping how Alexa skills get built. If you've shipped voice experiences before, it's a useful read on how the platform's assumptions are shifting and what the next generation of Alexa apps can do.

📰 How Zoox Designs Audio With Intent

Sound is one of the hardest interfaces to get right in an autonomous vehicle, and Zoox digs into how its team designs audio as a deliberate signal rather than an afterthought. The post covers how the robotaxi uses sound to communicate intent both to riders inside the cabin and to pedestrians and cyclists around it, and why every cue has to be legible without being alarming. If you build interfaces where a system has to convey state to people who never chose to interact with it, the reasoning here about clarity, restraint, and trust transfers well beyond a car.

📽️ AWS Partners with Anthropic on the Model Hardware Standard

Cagatay Cali and Albert Zhao unpack the Model Hardware Standard (MHS), an open standard that lets agents plug into hardware the way MCP did for software, with a humanoid-robot demo (Neon) to show it working. AWS (via Strands Agents) and Anthropic are building the standard together and testing it against real robots, and a research preview is open for access requests. If you're wiring agents to anything physical, this is the emerging interface layer to watch, and there's a Discord to follow development.

📰 The best tech products of 2026, according to USA TODAY

By Marc Saltzman

USA TODAY's year-end roundup features several Amazon devices, led by the Ember Artline with Fire TV, a 4K set that leans on three Amazon platform bets at once: the LLM-based Alexa+ assistant, an AI "Match the Room" feature that scans a space from your phone and generates ambient art to fit it, and Omnisense presence detection that wakes the display when someone walks in. The Ring Smart Smoke + CO Alarm also makes the list. For developers, it's a useful read on where Amazon is putting on-device and generative AI in ambient hardware, and on the sensing and personalization surfaces those experiences are built on.

📽️ Automate your code and tasks with Kiro Crew

Erik Hanchett demos Kiro Crew, an open-source persistent agent workspace that runs on your own hardware (Mac, Linux, Windows) rather than in the cloud. The demo covers the whole loop: installing with a single curl command, running setup, chatting from the terminal, wiring it into Discord, and handing it a real task, plus how it accumulates memory across sessions. It started as an internal Kiro tool and is now open source with hundreds of contributors, so if you want a self-hosted agent you can extend and keep your context on your own box, the GitHub repo and docs at kiro.dev/docs/crew are the place to start.

📰 Prime Video Uses AI to Sync Actors' Mouths in Dubbed Titles, Starting With 'Maxton Hall'

By Todd Spangler

Prime Video is rolling out AI-driven lip-sync that reshapes an actor's mouth movements to match the dubbed audio track, debuting on the German series "Maxton Hall" before expanding to more titles. For developers, it's a concrete look at generative video applied at streaming scale, where a model has to alter faces frame-by-frame while holding identity and expression stable across a full season. If you work on media pipelines or localization, it also reframes dubbing as a rendering problem rather than an audio one, with the quality bar set by whether viewers ever notice.

📰 IMDb Adds Digital Creator Profiles for the First Time

By Todd Spangler

IMDb is extending its catalog beyond film and TV to include dedicated profiles for digital creators, the first time the database has formally recognized online-native talent. For developers, it is a reminder that IMDb's entity model and public data surface keep widening, which matters if you build anything against IMDb metadata or reason about how Amazon structures entertainment identity data. Worth a look at how a decades-old catalog schema absorbs a category it was never designed for.

📰 Reducing cost and improving performance with Claude Platform

By Lance Martin

If you're building on Claude, this post lays out the three levers that cut your token bill without degrading output: prompt caching to avoid re-sending stable context, tightening instructions so the model does less work per call, and dialing effort to match the task. Each is a knob you can turn today in your own application code, and the guidance is concrete about when each one pays off. Worth a read before your next round of API cost tuning.

📬 Feedback

Questions or comments? Have an item for next week's post? Let me know!