Welcome, AI-curious product people.

Hey everyone 👋, thanks for reading Product Prompts, where I take the week's biggest events in AI and try to go beyond the hype, getting under the skin of what's happening from a Product perspective. If you're reading this and haven't signed up, join the 2,593 others by clicking Subscribe, and to the regular readers, thank you. 🙏

In today’s Product Prompts:

📣 Soapbox: Are we witnessing the birth of the post-smartphone experience? Rabbit R1 and the implications for product teams.

👀 AI startups to watch: Mistral AI, the new French LLM unicorn.

🧠 Things to know: ChatGPT can learn from your chats; the new ChatGPT Teams plan; the GPT Store is live.

🛠️ AI Toolkit: Olly - your AI-powered social media assistant; Kusho - Never test your APIs manually again. Marblism - Launch your React and Node.js app in minutes.

📖 ReadMe: A Rough Guide to the Rabbit r1; Gen AI (probably) won’t take your job, but it will change how you work; Vodafone to use Microsoft generative AI after signing £1.2bn deal.

Read time: 6 minutes

📣 WEEKLY SOAPBOX

📣 Are we witnessing the birth of the post-smartphone experience?

Last week we were introduced to a new AI-powered device, the Rabbit R1. And back in November, Humane launched its Pin, an AI-powered wearable.

Both are powered by OpenAI. The Rabbit R1 sold out on day one - a testament to the market's readiness for novel (and practical) AI-powered devices. Whereas Humane has seen a rather less positive reception. (If you want an example of how not to launch your product, watch their launch video.)

With the Rabbit R1, it's not just the features that have captured the early adopters’ imaginations but also the underlying promise of a more intuitive, less intrusive tech experience. Unlike smartphones, which have become synonymous with endless scrolling and distraction, Rabbit R1 champions a more focused, task-oriented approach. 

The R1 watches what people do on their screens, listens to what they say while they do it, and then, with an ingenious “teach mode”, can encode that into repeatable actions.

Priced at $199, the device integrates a 2.88-inch touchscreen, a rotatable camera, and an innovative scroll wheel, powered by the Rabbit OS and the Large Action Model AI platform. I think it's poised to redefine how we approach everyday tasks.

📣 I'm willing to bet that this is just the beginning of a broader shift towards devices that augment our digital lives without overwhelming them.

The R1 operates on a rather ingenious principle: it keeps on-device processing to the bare minimum, focusing primarily on natural language recognition. The brains of the operation lie in its “Large Action Model” (LAM), a clever bit of tech that taps into a variety of services to execute the user's commands.

Most of the heavy-duty work is smartly 'farmed out' to third-party apps and web-app APIs, including advanced gen AI services like ChatGPT. Users link these services to their Rabbit R1 through the aptly named 'Rabbit Hole' web portal, creating a seamless hub of digital activity.

On the other hand, Humane takes a different route, functioning like a wearable, AI-powered search engine. It's not just about information access; it's about rethinking communication and interaction. This device hints at a future where our digital assistants are not just in our pockets but part of our daily wear, offering seamless and intuitive assistance​​.

Humane's journey to market with its AI Pin illustrates the hurdles of bringing revolutionary products to market. Despite some initial pre-launch excitement, the recent layoffs, amounting to 4% of its workforce, and the reshuffling of key positions, including the CTO transitioning to an advisory role, paint a picture of a company grappling with internal challenges and external skepticism​​​​​​. This development, especially so close to the product's launch, raises questions about the company's strategy and market positioning. The high price tag of $699 and $24/month didn’t exactly help.

It's a classic case of vision confronting the hard realities of execution and market acceptance.

So, 2 AI-powered devices with 2 very different traction stories so far. But we should be ready for an increasing number of startups attempting to displace smartphones as the default personal piece of tech, and I believe we’re entering a new era of interaction for which product teams need to prepare.

🤔 We may not be ready for expensive AI wearables, but we are definitely intrigued by affordable pocket companions that automate a lot of the tasks we use traditional apps for.

Here’s my take:

💬 Voice-first interfaces are the a future. The transition to devices like Rabbit R1 signifies a move towards voice-first interfaces. This shift is set to challenge and inspire product teams to innovate beyond the traditional touchscreen paradigm​​.

💬 Simplicity and accessibility define the New Tech Era. The simplicity and accessibility provided by devices like Rabbit R1 will push product design towards inclusivity. This approach will especially benefit users who find current smartphone interfaces complex or challenging to navigate​​.

💬 Compact and wearable devices redefine portability. The varied form factors of AI devices like Rabbit R1 and Humane AI Pin represent a significant shift from the bulky smartphones. I can envision a future where the need for carrying a large device is replaced by more convenient and varied gadget options​​.

💬 Privacy-centric design is a must for user trust. With devices like Rabbit R1 requiring access to various services and personal data, prioritising privacy in design becomes imperative. I anticipate a growing focus on secure data practices as a crucial aspect of tech product development​​.

💬 Smartphones face a new breed of competitors. Despite the advancements of smartphones, AI devices like Rabbit R1 and Humane AI Pin bring unique experiences to the table. It will be fascinating to see if they can carve out a niche in a market dominated by smartphones​​.

💬 Now we need to design websites and apps for AI users, not just humans. This is a much wider topic than just being related to new physical AI devices, as ChatGPT and its ability to support “actions” (allowing GPTs to use 3rd party services to retrieve information or take actions) is already driving this phenomenon. I’ll cover it in detail in a future edition of Product Prompts, but for now consider the implications for user authentication, security, SEO, analytics, user metrics…

👀 AI STARTUP TO WATCH

👀Mistral AI

Mistral AI has made headlines for securing a significant funding round of $415 million. This impressive funding has boosted the company's valuation to around $2 billion, an extraordinary feat for a company that's only 7 months old​!

What makes Mistral different from its larger competitors is its open-source philosophy, an approach that fuels competition with industry giants, holds inherent regulatory challenges, and raises questions around the broader implications of its approach (open source = harder to regulate = potential misuse).

Mistral AI's core product is its large language models, with Mistral 7B being the first model released. Although smaller in size compared to competitors like GPT-4 and Gemini, Mistral claims superior efficiency and cost-effectiveness in processing and generating responses. The latest model is Mistral Medium, their most powerful model. It supports a context window of 32k tokens (around 24,000 words).

It’s arguably the closest non-OpenAI model to GPT-4 / GPT-4 Turbo, and is available, for you, for free, in chat mode, to try on: http://labs.perplexity.ai (and I wrote about Perplexity being a startup for product people to watch, last week.)

🧠 THINGS TO KNOW

🧠 Your GPTs can now learn from your chats. Access is being rolled out gradually and I don’t have it yet, but here’s a screenshot of what’s coming:

It's probably using some form of vector storage, which is how LLMs are working with extended memory. So it shouldn’t really impact the context window - it's like an external database call.

🧠 The OpenAI GPT Store launched (it’s free to view the GPTs but you’ll need ChatGPT Plus to use them). I covered the implications of this big bet from OpenAI for product teams in last week’s Product Prompts.

🧠 A ChatGPT Teams plan is now available, $30/month or $25 if paid annually. I can see this being super popular:

• message cap per user is 100 messages per user / 3 hours (vs 40 messages / 3 hours on ChatGPT Plus plan)

• No training on your business data or conversations

• Secure workspace for your team

• Create and share custom GPTs with your workspace

• Admin console for workspace and team management (i.e. to to configure permissions for everyone?)

🛠️ AI TOOLKIT

📖 README

Rough guide to the Rabbit R1 - link

Generative AI (Probably) Won’t Take Your Job. But It Will Change How You Work - link

Vodafone to use Microsoft generative AI after signing £1.2bn deal - link

🎬 THAT’S A WRAP 

THE AI ADVANTAGE

Every Sunday, I’ve started a new edition of Product Prompts called The AI Advantage, where a stellar guest from the product world shares the AI-powered tactics and prompts they use to supercharge their workflow.

Check out last week’s edition featuring Adelynne Chao, founder of Untold Insights, sharing how OpenAI helps drive customer-led discovery.

And if you’d like to be a future guest, reply to this email and I’ll be in touch!

FEEDBACK

If you have specific feedback on this newsletter, or anything you’d like to share, let me know by replying to this email.