James Perkins

Cofounder of What About Ai, Ex-JPMC, tech junkie, family man, & practical environmentalist. Founder Tek Perks, Co-Founder CityRyde. All views my own.

[The Babylon Bee @TheBabylonBee]

'Oh, Your Intelligent Creations Are Trying To Kill You? That's Interesting,' Says Jesus https://t.co/3fUeB5qbGs

[James Perkins @jamesrperkins]

Anthropic Claude Fable 5.1 is releasing today (after a brief delay due to technical difficulties)!

I’m hearing enhancements include fewer false trips of safeguards, better performance at lower levels of effort akin to Fable 5 at Max, and improvements to Cowork aligned to recent Grok Bot experience.

[James Perkins @jamesrperkins]

There is a need for an indoors play space between the age range of the Please Touch Museum and the Franklin Institute. If they turn the remainder of the whole 3rd floor into an expanded playroom with climbing implements and play activities targeting ages 4-10 they will get a ton of membership, and visits in the cooler months.

The butterfly hatchery was also big.

The stuffed animal displays don’t draw crowds any longer, but the museum is an awesome part of the city and deserves to remain open to the public.

[James Perkins @jamesrperkins]

@AvitalBalwit @_sholtodouglas Finding someone you enjoy being with and doing nothing, just as much as going on an adventure, and being able to laugh together unadulterated… that’s where it’s at. Congratulations to you both!

[James Perkins @jamesrperkins]

@leopoldasch The sincere joy captured in that moment is beautiful (and so very human). I wish you both continued happiness and growth together! Congratulations! 🍾

[James Perkins @jamesrperkins]

Contact me if you’re looking to see similar results in your business.

Companies need to focus on using ai effectively/efficiently, not limiting tokens.

[Praveen Neppalli @praveenTweets]

As our CFO @_balaji_km mentioned at earnings today, we’re seeing some very interesting trends on AI costs. I think it’s another signal that we’re coming to the end of the so-called ‘tokenmaxxing’ era.

Here’s what’s been happening behind the scenes.

Since the beginning of the year we’ve more than quadrupled the number of people using frontier AI tools. That’s thousands of engineers using them every single day. During that same period, our cost per token has declined.

You might expect costs to rise as adoption accelerates. We've seen the opposite. Not because we've restricted access, but because we've treated efficiency as an engineering problem rather than a budget problem. A few examples:
• Caching and reuse: We use optimizations to improve our prompt cache hit rate that reduce our input token spend.
• Better defaults and tooling: We tuned default model settings, context sizes and developer workflows so teams get the same results with fewer tokens and lower-cost inference.
• Visibility drives efficiency: We gave engineers real-time visibility into their AI usage and costs per hour.
• Experimenting with open-weight models: we continuously evaluate new models and deploy the best option for each use case.

This is the future of applied AI at enterprise scale. The next phase, whatever we call it, will not be characterized by who spends the most tokens, but about how people use them as efficiently as possible.

Credit to all the engineers at @Uber who are helping to build this future. 🚀

[James Perkins @jamesrperkins]

@stephen_wolfram A beautifully written tribute, and great comfort in knowing that Elise’s joy-filled day was the immediate precursor to her swift death. May the many memories you’ve shared ease your pain, and help your family through Elise’s passing.

[James Perkins @jamesrperkins]

This is very likely a vengeance-based bargaining chip deployed before the govt taking a stake in the AI frontier labs. It’s “give us what we want, else be nationalized, or worse, handicapped/shut down”

[Dave Blundin @DaveBlundin]

This is one of the most pivotal events in history because of the precedent it sets. Even if it doesn't stick, the cat is out of the bag: the government can now unilaterally decide who gets to use AI.

Over the last few days I had fully transitioned to orchestrating with Fable. It initially felt like the cookie jar was moved out of reach. Now after spending the day with Opus, I've accepted just how monumental this is.

Being denied the best AI model is effectively the same as being denied future employment.

[James Perkins @jamesrperkins]

Mythos level models are coming in the next few weeks, and for now, a sizable increase in agentic coding with release of 4.8

[Claude @claudeai]

Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors.

Available today at the same price.

[Claude @claudeai]

Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors.

Available today at the same price.

[Rob Wiblin @robertwiblin]

When I interviewed Holden Karnofsky he gave a plausible justification for why Anthropic should remain at/near the AI frontier. ¹

But the case is far weaker for running far ahead of the competition on approx the most dangerous capability: autonomous SWE agents that can set off a recursive self-improvement loop.

These capabilities are also now being stolen by less scrupulous actors: https://t.co/FbvBacIT0e

As Holden said: "If AIs were able to just do AI R&D I think there would be a significant chance of what I tend to call a capabilities explosion.

And then everything else you're worried about from AI, all the other threats... all that comes onto the table really fast, and you're not going to have time to react to it."

Full interview here: https://t.co/9IrFREbfhM

¹ If you believe technical alignment is very difficult and none of the current approaches have much of a chance I agree the case for this whole strategy is much weaker.

[Claude @claudeai]

Another Claude Code hackathon comes to an end.

Thank you to everyone who spent a week building with Opus 4.7, and to @cerebral_valley for co-hosting.

Introducing the winners:

[James Perkins @jamesrperkins]

This highlights the feverish race for Ai dominance underway, & the lengths with which the Chinese govt will go to ensure they come out on top.

Think twice about the details you feed into Deepseek and other Chinese models if you’re in the US and working in or near a frontier lab.

[White House OSTP 47 @WHOSTP47]

https://t.co/xjjjvNaAIz

[James Perkins @jamesrperkins]

Dave is my favorite present-day futurist talking about Ai & the singularity. Always a thoughtful, centered, & experience-based takes with passion and vision for best case scenarios. All co-hosts of Moonshots are great, too, but my ears perk up when Dave speaks.

Check him out!

[Dave Blundin @DaveBlundin]

ANNOUNCEMENT!!!! A lot has been happening in the singularity, and I realize many of you want me to share more of it. You've been asking for more behind the scenes, more founder content, and more strategic and actionable content beyond Moonshots.

So, big news! I'm officially expanding. My team and I will be working to bring you more of these types of content and you can now find me on Instagram, TikTok, YouTube and Substack. Follow along and let us know what you think. We're just getting started!

[James Perkins @jamesrperkins]

Nearly no one understands how powerful the existing Ai tools are, nor how fast they’re changing.

You have 6-18 months to do something meaningful with a paid subscription before the frontier labs start locking down public access.

[Dave Blundin @DaveBlundin]

Sam Altman said @OpenAI would deliver 100x cheaper intelligence by end of 2027. I said we'd get there in one year, not two.

@Google's TurboQuant breakthrough compresses AI memory usage by 6x during inference, but everyone's missing the real story. It's not about running models on your phone. It's that the smartest AIs can now hold more context, or use 8x more thinking tokens, or you can train 8x larger models. We'll use some combination of those to get the best results. And this is just one of a dozen compounding breakthroughs. AI is helping improving AI, and that loop is accelerating. At this rate, we might blow past 100x by summer!

[Dave Blundin @DaveBlundin]

Sam Altman said @OpenAI would deliver 100x cheaper intelligence by end of 2027. I said we'd get there in one year, not two.

@Google's TurboQuant breakthrough compresses AI memory usage by 6x during inference, but everyone's missing the real story. It's not about running models on your phone. It's that the smartest AIs can now hold more context, or use 8x more thinking tokens, or you can train 8x larger models. We'll use some combination of those to get the best results. And this is just one of a dozen compounding breakthroughs. AI is helping improving AI, and that loop is accelerating. At this rate, we might blow past 100x by summer!

[James Perkins @jamesrperkins]

@SpaceX Here’s to 600 more next year! Thank you for forging the way by driving innovation that’ll carry humanity and our technology into the future, and simply as the next step in a greater 100x plan.

We’re so fortunate to be living at this time in history. Congratulations!

[James Perkins @jamesrperkins]

@JeffBezos @Trace_Cohen Congratulations! There’s not much of a second mover advantage when it comes to landing on a point in the ocean. This is a huge achievement for humanity, and for the whole team involved.