😺 10 EPIC AI voice chat demos

Welcome, humans.

Someone made a compilation of the hot new video model MiniMax’s hot new image-to-video feature. The results? This, you guessed it, “hot new” sizzle reel—seriously though, the quality of these clips is outrageous:

Looks like Minimax is worth the upgrade

The creator says workflow here is Midjourney to MiniMax, with soundtrack from Udio.

Here’s a full tutorial video that compares the MiniMax output to other models, and here’s a quick snapshot of the top AI video tools out there.

Here’s what you need to know about AI today:

• We share 10 wild OpenAI voice app demos.
• OpenAI thwarted over 20+ attempts at election meddling.
• Walmart introduced its new retail AI models.
• OpenAI to wide release SearchGPT before 2025.

Advertise in The Neuron here.

10 mind-blowing OpenAI realtime API examples that'll make you say “Hey Siri, you're fired.”

It’s been less than a week since OpenAI announced its realtime voice API, and developers are going wild. This tech lets apps have natural, real-time conversations with users, making those awkward pauses with Alexa a thing of the past.

Want to see hear what the future sounds like? Check out these 10 examples:

1. Camera bot: Dr. Bobby Gomez-Reino engineered a voice controlled tour of his virtual data center, where he changes camera angles by chatting with his bot.
2. Browser whisperer: Sawyer Hood built a voice-controlled web browser. “Google, show me cat videos" just got a whole lot easier.
3. Speech to Picasso: Jordan Singer splashed together a voice-controlled painting app.
4. PDF mind reader: Marcus Schiesser created a voice chat for documents. “Hey term paper, what's your main argument?” Yes, please.
5. 5-minute assistant: Pietro Schirano whipped up a voice assistant with Claude in “one shot.”
6. Interview prep pal: Kenn Ejima prepared an AI interviewer to conduct a 2 minute mock interview, quizzing you on your resume experience.
7. Smart voice agent: LangChain, an AI agent developer, crafted a voice assistant that can use tools like a calculator (code).
8. Website dialogue: Nicolas Camara made it possible to chat with any website (like get the latest headlines from Hacker News, for example).
9. Stock tracking assistant: Willy Douhard made a voice assistant that can chart the price movements of multiple stocks with only your voice.
10. Real time animated friend: Bryan Pratte shared how to combine OpenAI’s voice AI with ExpressionEngine to bring his animated characters to life.

More examples here.

Where this is all going: Replacing humans in customer service, of course! For example, Sierra, co-founded by ex-Salesforce exec Bret Taylor, is raising billions to disrupt customer service with AI that handles phone calls.

With 100K+ call centers employing 17M globally, VCs are salivating over the potential of 24/7, burnout-free AI support. No wonder Sierra's valuation might triple to $4B.

FROM OUR PARTNERS

💼 Want to build a 6-figure business as an AI Consultant?

This stat is WILD: the AI consulting market is about to grow by a factor of 8X to $54.7B in 2032.

So how do you, an AI enthusiast, become a handsomely compensated AI consultant?

Our friends at Innovating with AI have the answer:

The AI Consultancy Project, a new program that trains you how to build an AI consulting business. And they’ve already welcomed 300 new students!

Current students love:

• Tools + frameworks to find clients and deliver top-notch services.
• A 6-month plan to build a 6-figure business.
• Getting their first AI client in as little as 3 days.

As a reader of The Neuron, you can click here to request early access to the next enrollment cycle!

Around the Horn.

Here’s the project page for more.

• Anthropic published their policy approach to detect and mitigate election misinformation, addressing the fact that Claude doesn’t produce image, audio, or video to avoid deepfakes—this caused some speculation that Claude 3.5 Opus will only be released sometime after the election (for safety reasons).
• OpenAI has disrupted 20+ attempts by bad actors to use its models for election interference across multiple countries.
• Walmart got into the language model game with Wallaby, a suite of retail-focused models that are trained on “decades” of the company’s data—the project is being tested now, and will roll out over the next year to consumers.
• SearchGPT, ChatGPT’s rival search engine to compete with Google and Perplexity, will be rolled out to ChatGPT “by the end of the year.”

Treats To Try.

1. *"I'm blown away by the high quality and value of this event." - Ricardo B.
2. "Great event - worth getting up at 4am in the morning for!" - Sandy A.
3. "I loved the presentations and was truly captivated by the depth of experience and insight shared on these panels!" - Peter K.
4. Don’t miss GenAI Productionize 2.0—the premier conference for GenAI application development, featuring AI experts from leading brands, startups, and research labs! Register for FREE here.
5. PyramidFlow is a new open-source video model that generates 10-second videos at 768p resolution from your text prompts or images (paper here, read more here, and demo it here).
6. HuggingFace launched Gradio 5, which lets you build AI apps in a few lines of code, now w/ instant loading and real-time streaming (demo here).
7. Cooraft transforms your selfies and photos into studio-quality videos or artistic animations (iOS / app store only rn, as its a newly launched tool. We haven’t tried it yet, so check out this demo video here and see if its for you).
8. Blocfree lets you create and share email templates with just a few clicks, then publish them across multiple platforms (free to try, its pretty fast to test out).
9. Latitude helps you build, test, and improve your prompts when using AI in your own app and need to format the code properly (video demo here).
10. Height 2.0 automatically tracks project progress, writes updates, and cleans up your task lists for you (fun interactive demo when you sign up).
11. Coverr teaches you the workflows for how to make select AI videos using video models like Midjourney and Runway.

See our top 51 AI Tools for Business here!

*This is sponsored content. Advertise in The Neuron here.

Intelligent Insights.

Can OpenAI's ChatGPT Beat Humans At Gaming?

• DeepMind's Michelangelo test shows that even the most advanced long-context reasoning models, capable of processing millions of words at once, struggle with complex reasoning tasks over long contexts.
• OpenAI introduced a new benchmark to compare AI agents against humans (*cough* agents are coming *cough*), called MLE-bench, which uses 75 real-world challenges to test the agents against human experts—and its best AI system (o1 + AIDE) earned top marks in about 17% of these contests (paper).
• Check out this paper from Cornell researchers trying to make AI smarter at finding info: they used “contextual document embeddings” to help AI grasp document context, boosting accuracy in specialized fields by catching subtle differences regular systems miss (here’s a small version you can try).

A Cat's Commentary.