Lessons From My First AWS re:Invent

IN THIS ISSUE: My conference adventures continue as I journey to Amazon’s re:Invent conference to hear about AWS’ approach to AI. Plus, explore the company’s AI partnership with the NFL and how they use the technology to help support players’ philanthropy. And don’t miss the roundup of AI news you may have missed from the past week!

The Prompt

Last week, I found myself in the bustling heart of Las Vegas, attending Amazon Web Services’ re:Invent developer conference for the very first time, courtesy of Amazon. The anticipation was high as I waited to hear about the company’s new AI innovations.

Over four jam-packed days, the conference delivered a flood of announcements. It was easy to feel overwhelmed! As I navigated through prongs of attendees and trekked across sprawling hotels to catch keynotes and meetings, I struggled to process everything AWS unveiled. But eventually, I made sense of the news, and I’m happy to share what I’ve learned from re:Invent.

Disclosure: I attended Amazon's 2024 re:Invent as a guest, with a portion of my travel expenses covered by the company. However, Amazon had no influence over the content of this post—these thoughts are entirely my own.

Amazon Nova

One of the more surprising announcements came from Amazon Chief Executive Andy Jassy during the day one keynote. The company unveiled a foundation model named Amazon Nova. Integrated into Amazon Bedrock, it’s billed as low-cost, low-latency, and customizable. Available in six variations—Micro, Lite, Pro, Premier, Canvas, and Reels—Nova is a proprietary model to take on Google, Meta, OpenAI, and even Anthropic.

However, Amazon doesn’t believe that one model will rule them all. That’s why it has invested more than $8 billion into Anthropic. The new foundation model is part of AWS’ efforts to give developers a choice, making dozens of models—Nova being one of them—available within its Bedrock platform to build their own intelligent applications.

Amazon has more than just this foundation model in its arsenal. The company boasts a range of homegrown models, including Titan, along with several other specialized options.

One differentiator I’ve learned about Amazon Nova is that it’s a model Amazon has been using internally across multiple business verticals. From its Rufus smart shopping assistant and Prime Video to Amazon Pharmacy, the e-commerce giant says it has established solid business use cases for Nova. This demonstration could appeal to business leaders, something many leading model makers can’t readily do.

Read more: What you need to know about Amazon Nova, its six variants, and how the company believes it differs from other proprietary models today.

Don’t Sleep on Inference

Inference is how an AI system applies what it has learned to make decisions or predictions with new information. AWS Chief Executive Matt Garman calls it a vital part of every intelligent application—and that’s exactly where Amazon Bedrock comes in. To support that, the company introduced a slate of new features to the development platform designed to aid in producing AI apps. At re:Invent, AWS introduced automated reasoning, model distillation, multi-agent collaboration, along with prompt caching and better data integration.

In order for an AI to make better decisions, it needs better information. And the new slate of tools in Bedrock make it possible for AI to perform more accurately. It’s “where customers can build inference-oriented applications, and that’s where they really see value from AI,” AWS’ Vice President for AI and Data, Rahul Pathak, tells me. “Because it’s focused on driving customer value, it’s really important for customers to understand what that is. And the space is moving super fast. That’s why there’s a ton of innovation happening. We believe, in order to build great gen AI applications, you need a few key pillars: Choice of model; access to data, which is the knowledge base; ability to scale in front; responsible AI guardrails; and then you need agents. That’s how you ultimately build an inference-oriented application that can drive value.”

Supporting the Next-Generation of AI Startups

AWS has a long history of supporting startups, and it’s no different when it comes to gen AI. The company has set up numerous AI accelerators worldwide to foster companies looking to tap into the Amazon ecosystem. Two years ago, it expanded this program with the launch of a global effort, inviting select startups looking for more cross-border exposure. Out of the thousands that applied, 80 were ultimately selected.

I was fortunate to meet with some of these participants at the conference, some of whom aren’t using gen AI for content creation but for robotics, helping visually impaired people see, and more.

It all culminated with the inaugural Unicorn Tank Pitch Competition, a “Shark Tank”-like event in which eight startups battled for $100,000 in prize money. Presenters had three minutes to pitch the panel of five judges, including Aileen Lee, the investor who first coined the term “unicorn.”

I’ll have more to say about this generative AI accelerator, or GAIA, in a future blog post. Nevertheless, after seeing the presentations and interviewing Tiffany Bloomquist, AWS’ Head of Startups for Asia Pacific and Japan, I feel more informed about how support and the credits like what AWS provides startups can be a critical lifeline, especially when dealing with the resources needed to operate such technology.

“One of the things that we know about being a startup from our early days is that…it’s important that you’re thinking big, but you are starting small, and then scaling as fast as you can. That is a mental model we have at Amazon,” Bloomquist says. “It’s something we’re very proud of, but often that experimentation in the early days can be very expensive. And so, one of the ways that we want these scrappy startup teams to be able to experiment at pace is by creating programs that have to support them in their growth.”

Featured Image: Ken Yeung awaits a press conference at AWS' re:Invent 2024.

---

How AWS and the NFL Are Using Gen AI to Help Players Support Their Favorite Causes

“My Cause My Cleats” is an annual effort by the National Football League to raise awareness for players’ favorite causes. Athletes would don specially created cleats to support those efforts, though typically, local artists would handle the creation.

As part of its renewed deal with the league, AWS is lending its AI technology to help several star players design their shoes. Through text and natural language, Buffalo Bills quarterback Josh Allen, Seattle Seahawks wide receiver DK Metcalf, and Las Vegas Raiders defensive end Maxx Crosby created their cleats. To make the design feel more connected to their charities, Allen (John Oishei Children’s Hospital in Buffalo) and Metcalf (Prison Fellowship and SOUND) collaborated with the people behind the causes on the design.

The creation was done through an app powered by Anthropic’s Claude 3.5 and Stability AI models. Players controlled the shoe’s appearance, style, colors, and texture. When complete, a local artist handled the physical production.

To broaden the reach of “My Cause My Cleats,” AWS launched an interactive website for anyone to make their own cleats. You can be as specific or generic with your design prompt. Choose from ten different visual styles and regenerate the image until you’re content with the look and feel. Unfortunately, neither AWS nor the NFL will mail the physical shoe to you, although nothing stops you from taking the design and, on your own, having a shoemaker do it for you.

Read more about this partnership and how fans can win the game-day cleats from one of the players.

---

Today’s Visual Snapshot

A look at the landscape of AI agents as documented by AIAgentsDirectory.com as of September 2024. The site catalogs 258 agents servicing many different fields, from agent building and coding to productivity, customer service, data analysis, personal assistance, research, content creation, and more.

---

Quote This

“[G]enerative AI inference is going to be a core building block for every single application. In fact, I think generative AI actually has the potential to transform every single industry, every single company out there, every single workflow out there, every single user experience out there […] I think inference is going to be part of every single application […] Every application is going to use inference in some way to enhance or build an application. And if you’re going to do that, it means you need a platform that can deliver inference at scale.”

— Amazon Web Services Chief Executive Matt Garman at the company’s re:Invent conference last week.

---

This Week’s AI News

🏭 AI Trends and Industry Impact

• The GPT era is already ending (The Atlantic)
• What outgoing White House chief technology advisor Arati Prabhakar has to say on AI (MIT Technology Review)
• Google CEO Sundar Pichai: AI development is finally slowing down—”the low-hanging fruit is gone” (CNBC)

🤖 AI Models and Technologies

• OpenAI’s active user count soars to 300 million people per week (CNBC)
• OpenAI launches full o1 model with image uploads and analysis, debuts ChatGPT Pro (VentureBeat)
• Meta unveils Llama 3.3 70B, a new, more efficient model (TechCrunch)
• Ai2 unveils OLMo 2, calling it the most advanced fully open language model yet (My Two Cents)
• Microsoft’s Copilot Vision is here, letting AI see what you do online (VentureBeat)
• OpenAI charging $200 per month for an exclusive version of its o1 “reasoning” model (The Verge)
• China’s generative AI users reach 230 million as startups, Big Tech roll out LLM services (South China Morning Post)
• DeepMind’s Genie 2 can generate interactive worlds that look like video games (TechCrunch)
• Luma AI launches Ray 2, a model enabling video creation from text and images in seconds (My Two Cents)

✏️ Generative AI and Content Creation

• OpenAI releases Sora, its buzzy AI video-generation tool (CNBC)
• ElevenLabs launches feature to create and edit AI-generated podcasts (Bloomberg)
• Google Cloud launches Veo AI video generator model on Vertex (VentureBeat)

💰 Funding and Investments

• The creator of ChatGPT’s voice wants to build the tech from “Her,” minus the dystopia, raises $40 million (TechCrunch)
• Elon Musk’s xAI lands $6 billion in new funding to fuel AI ambitions (TechCrunch)
• Why investors don’t mind that AI is a money pit (The Verge)
• Cleerly raises $106 million from Insight Partners for AI heart health early detection (TechCrunch)

☁️ Enterprise AI Solutions

• Amazon CEO Andy Jassy reveals AWS’ strategy for building the enterprise AI platform (SiliconAngle)
• AWS looks to conquer cloud complexity with “simplexity” (ITProToday)
• AWS makes SageMaker HyperPod AI platform more efficient for training LLMs (TechCrunch)
• The secret weapon helping businesses get results from AI: Humans (The Wall Street Journal)
• The future of AI agents: Highly lucrative but surprisingly boring (The Financial Times)
• Qodo’s fully autonomous agent tackles the complexities of regression testing (VentureBeat)

⚙️ Hardware, Robotics, and Autonomous Systems

• AWS’ Trainium2 chips for building LLMs are now generally available, with Trainium3 coming in 2025 (TechCrunch)
• You can now try Microsoft’s Recall AI feature on Intel and AMD Copilot+ PCs (The Verge)
• How Samsara’s AI for the unglamorous could drive productivity gains (Semafor)
• AI-powered robots can be tricked into acts of violence (Wired)

🔬 Science and Breakthroughs

• How AI monitoring is cutting stillbirths and neonatal deaths in a clinic in Malawi (The Guardian)
• Google’s DeepMind releases weather prediction AI called GenCast, says it outperforms the world’s top weather forecast system (TechCrunch)

💼 Business, Marketing, Media, and Consumer Applications

• How Imogen Heap is using AI to shape the future of music (The Conversation)
• Key leaders behind Google’s viral NotebookLM have left to create their own startup (TechCrunch)

⚖️ Legal, Regulatory, and Ethical Issues

• Inside Britain’s plan to save the world from runaway AI (Politico)

💥 Disruption, Misinformation, and Risks

• AI friendships claim to cure loneliness. Some are ending in suicide. (The Washington Post)
• UCLA offers comparative literature class with textbook, homework assignments and teaching assistant resources generated by AI (TechCrunch)

🔎 Opinions, Analysis, and Editorials

• Will AI eat the browser? (Crazy Stupid Tech)
• The cognitive cost of AI (Fast Company)
• Hidden AI revolution: Why leaders must address covert adoption of new tech (GeekWire)
• When AI plays doctor: The crisis in automated health insurance (LinkedIn/David Linthicum)

End Output

Thanks for reading. Be sure to subscribe so you don’t miss any future issues of this newsletter.

Did you miss any AI articles this week? Fret not; I’m curating the big stories in my Flipboard Magazine, “The AI Economy.”

Connect with me on LinkedIn and check out my blog to read more insights and thoughts on business and technology. 

Do you have a story you think would be a great fit for “The AI Economy”? Awesome! Shoot me a message – I’m all ears!

Until next time, stay curious!