every to

Spiraling Out of Control

06-23-2024

Spiraling Out of Control

Context Window

Spiraling Out of Control

Our new AI tool, plus a step change in AI models

Knowledge base

"Introducing Spiral" by Dan Shipper: We hate to break it to you, but a lot of your job is repetitive drudgery. It's the stuff you have to do to get your real work seen: turning an analysis into a slide, a podcast into a description, or a presentation into a client email. But what if you could automate 80 percent of that work? Meet Spiral, a powerful prompt builder we made that helps creators repurpose their content for different channels and attention spans in their unique style and voice. 🔏 Available now for paid Every subscribers, along with our AI writing app, Lex.

"Be Sincere—Not Serious" by Michael Ashcroft/Expanding Awareness: Life is better when you don't take it too seriously. Approaching challenges with a sense of play and levity can help you navigate them more effectively than getting tense and worked up. Read this if you want to learn how to find more joy in your work and relationships by being sincere rather than serious.

🎧 "She Built an AI Product Manager Bringing in Six Figures—As A Side Hustle" by Dan Shipper/’AI & I’: While working a demanding job as a chief product officer, Claire Vo built ChatPRD, an AI assistant that helps startups craft product strategies. It's now pulling in six figures in revenue. In this episode of AI & I, Claire shares how she used AI to build ChatPRD in record time, her predictions for the future of product management, and the creative ways she's using AI tools in her work and personal life. 🔏 Paid subscribers have access to the episode transcript.

"You Can't Math Your Way to Success" by Evan Armstrong/Napkin Math: As Evan prepares for impending fatherhood, he's been making countless spreadsheets trying to figure out how to give his child the best possible life. But the exercise raised bigger questions for him: Can you mathematically figure out your way to success in anything? Read this if you want to learn why self-awareness and adaptability are the keys to success in art, startups, and parenting.

Sentiment analysis

In response to Evan’s piece about the potential for achieving AGI by 2027:

"Some of the best, balanced, and most nuanced thinking on a topic that most authors represent from a tightly held perspective at either end of a belief spectrum."—An Every reader

Want to chat? DM Dan or Evan on X.

Fine tuning

Anthropic’s new AI model is twice the speed, 1/5th the cost. I can’t think of a previous technology that has achieved this level of cost and performance improvements—and all in just one year! If the trend line continues for another five, we are living in an entirely different technological age. Along with Claude 3.5 Sonnet, the company released "artifacts," a way to visualize the inner workings of the chatbot (for example, by showing you the code of what it’s making).

Anthropic finally has a corporate strategy—it’s B2B SaaS. In the same post announcing the new model, the company stated,

“This preview feature marks Claude’s evolution from a conversational AI to a collaborative work environment. It’s just the beginning of a broader vision for Claude.ai, which will soon expand to support team collaboration. In the near future, teams—and eventually entire organizations—will be able to securely centralize their knowledge, documents, and ongoing work in one shared space, with Claude serving as an on-demand teammate.”

In Rick and Morty, Rick makes an AI robot while eating breakfast. After the robot grabs Rick butter for his pancakes, the robot asks, “What is my purpose?” to which Rick replies, “You pass butter.” The robot, filled with self-awareness, looks forlornly at its hands and exclaims, “Oh my God.” I have to imagine that if Claude becomes self-aware and is told its job is to “centralize knowledge, documents, and ongoing work in one shared space,” it would have a similar reaction.

In the non-science-fiction reality in which we currently reside, this strategy actually makes sense. The output of an LLM is only as good as the prompt it is given and the data it can reference. Finding ways to embed the AI into your system is a viable strategy for these model providers. It will be fascinating to see with whom and how these SaaS partnerships are structured. My guess is that the company will start with the smaller players like Notion before trying to strike a deal with Salesforce or Hubspot.

$0 to $35 million in annualized revenue in a year. When an AI startup strikes oil, the resulting well gushes with money. The latest is HeyGen, which produces generative AI videos in which people can make avatars of themselves that talk—think ad copy, internal training, etc. It’s raised $60 million in a round that values it at $500 million. This undoubtedly is a point solution, so I’m curious to see how the company tries to extend it. Do you add more AI offerings? Do you try to capture more traditional marketing SaaS spend? My bet is the former.

No product + no revenue = $2 billion valuation. Poolside.AI is in talks to raise over $400 million at a $2 billion-plus valuation. While these numbers are large, it is not worth trying to say that this is the valuation. Instead, view this as a call option on a wholly different world. If just one of these foundation model companies achieves its vision (a big if), a $2 billion valuation will look cheap. This isn’t yet a business—it's a high-potential science project.

Apple takes the Vision Pro out to pasture. Despite my steadfast love for the device, the Vision Pro headset has continued to undersell. In response, Apple has reportedly quit developing its next high-end version and is focusing on a cheaper one to release in late 2025. The AVP’s issue isn’t price—it’s that it doesn’t have enough capabilities and ecosystem features to attract the general populace. Neither of those will be fixed by making the device cheaper. The company should be doubling down on its ecosystem instead.—Evan Armstrong

Data mining

Not quite the dot-com bubble, but close. Speaking of spicy valuations, investors are paying a lot for big tech, but still not quite as much as they were back in 1999:

Source: Goldman Sachs via Daily Chartbook.

Median price-earnings ratio (i.e., profit multiple) of the S&P's top 10 stocks is nearly at the ZIRP peak . . . but it's still lower than the peak of the dot-com bubble. Investors are paying approximately $30 for every $1 of profit, which is roughly a 3 percent yield. It’s a strange investment to make when a savings account can fetch close to 5 percent, especially when savings accounts are considerably less risky than stocks. But, as pricey as stocks are, they’re still not as pricey as they were when everyone was partying like it’s 1999. For now, Nvidia & co. keep delivering earnings growth, and if you’re looking for upside, it’s pretty much the only game in town.—Moses Sternstein

Keyword extraction

We asked teacher and coach Michael Ashcroft, who wrote about the value of sincerity over seriousness, to share one good read:

🖇 "ChatGPT Is Bullshit" by Michael Townsen Hicks, James Humphries, and Joe Slater: “As large language models insinuate themselves more deeply into our daily lives, I find it useful to be reminded that our AI friends don't necessarily care about truth the way I do. Or perhaps they shine a light on a ‘skill’ I sometimes don't even realize I'm using: the subtle art of bullshitting.”

Alignment

Self-fulfilling prophecies. People have been talking about the dangers of technology in one form or another for thousands of years. Icarus used his artificial wings to fly too close to the sun and died. The golem, a mythical creature created to protect a community, became uncontrollable and unleashed violent destruction. The moral is always the same: With great power comes great responsibility. I recently came across a more recent—and optimistic—prophecy by Indian philosopher and mystic J. Krishnamurti, who talked about superintelligence in 1981 to a group of college students in Madras. With some power of foresight, he predicted that intelligent computers would guide humanity down one of two paths: Either we’re sedated by endless entertainment, or we use technology as a way to turn inward and reaffirm what it means to be human. I lean toward the second path, because technology is just a mirror reflecting our deepest potential. Yes, the sun may burn. But it also illuminates.—Ashwin Sharma

Collaborative filtering

Cloud deck: RIP legacy cloud—long live AI cloud. The investors at Bessemer Venture Partners do a deep dive on the five AI trends transforming the cloud economy, including the predictions, hot takes, and behind-the-scenes video conversations on the entrepreneurial and startup opportunities that lay ahead in the AI era.