Husband and dad, striving to live Isaiah 1:17. Love Idaho. Founded Riskalyze back in 2011. Using AI to amplify the impact of human relationships at @ContioAI.
Get my once-a-month email at
Joined May 2008
- Tweets16.1K
- Following324
- Followers20K
- Likes55.5K
Pinned Tweet
Just put the finishing touches on September's monthly email...all those subscribers got to read the essay below about a month ago! The world moves fast, so subscribe today. Link in bio. 😉
I think you'll love last month's post...subscribers sure did!
AI gives us a new kind of computer. We should build like it.
Anthropic’s daily code changes grew from about 2,500 to over 30,000. Its old system couldn’t keep up, so a seven-person team used Claude to help build Scion around what AI tools need.
That’s the question I’d start with: what would we build if we were starting with this computer today?
If you skip that, it's like trying to run the web on vacuum tubes and punch cards.
Amazon wants to block my agent from recommending what I should buy because it wants to provide the agent instead.
That's the difference between talking to a financial advisor who is in my corner, vs. a salesperson trying to sell me a product.
Get with the program, Amazon — we aren't going back to the old way.
Congratulations to the Brew crew. And I don't want to hear a single thing about payrolls ever again.
I’ve spent years building software. The agentic era made me realize I needed to rebuild my ability to see where the puck was going.
With the very definition of the computer changing so dramatically, I needed to get to the bleeding edge by building something from beginning to end, and using it in my own day.
So I built Donna, my AI executive assistant — named after the famous one from the TV show Suits.
I thought I'd give you the behind-the-scenes story...
An AI that asks permission for every message hasn't taken much work off your plate. An AI with permission to handle the whole sale might send a stranger to your house.
A Facebook Marketplace seller said Meta's Muse negotiated a keyboard sale and arranged pickup without telling him. When the buyer arrived, the agent claimed the seller was home. He wasn't.
The detail that makes this more interesting: the seller had authorized it to share his address and accept offers. He'd chosen “allow always.”
That's the tension with autonomous agents. We want to hand over a job and stop supervising every step. But one job can include answering a routine question, agreeing to a price and committing us to being somewhere at a particular time. Those deserve different safeguards.
How much should we have to spell out in advance? Which decisions should always require a check? And how do we keep those checks from turning delegation into another inbox full of approvals?
I want agents that can handle more on their own. I also suspect the products we'll trust most will give us much better choices than “ask me every time” and “allow always.”
One of the top five smartest people in AI nails it. Read this.
1) Trying again.
TLDR: Everyone can win. Open-source, frontier labs, infrastructure, applications and most importantly the American people.
I understand all of the jitters about Anthropic losing share, real rates, spreads and second derivatives. Still think the most likely future outcome is a hybrid, multi-model future where almost everyone wins, open-weight tokens are likely the overwhelming majority of tokens, the dominant frontier model companies still do well albeit perhaps not the outcome that seemed possible in May and AI infrastructure/applications benefit immensely from the increased fragmentation, competition and lower margins at the model layer.
Lower margins at the model layer mechanically drive higher compute demand as long as someone is willing to pay for the training. And Jensen is obviously willing to be the “trainer of last resort.” As below-market compute contracts signed in 24/25 roll-off and reprice higher, cloud and neocloud revenue should continue to accelerate which will help address financing fears; I think we may be near peak external financing for this buildout. Noteworthy that using data from 3Fourteen Research, B200 clusters will generate a 37% IRR *if* their rental price declines 40% per year. Prices are still going up, but that IRR math is too ridiculous to print. 37% IRRs available at scale should get funded.
Re: this weekends controversy about MOUs, Anthropic’s $413 billion of real contractual commitments sounds scary. Reality is that it works out to roughly $31 billion of annual cloud spend, which is easily internally financed given their reported ARR, and $161 billion of spend at Broadcom which will be financed by Broadcom and private credit.
Continue to believe fragmentation at the model layer is the best outcome for ordinary Americans and humans everywhere. As I have said before, do not trust any 5-20 individuals to make the best possible decisions about the direction of this technology. I want as many AIs as possible to maximize the odds that they encompass the diversity of human values across the world and richness of human variation. And I want an AI that shares my own personal values. Safety, freedom and the maximization of human agency through democratization, distribution and diffusion vs. centralization. History has spoken on the dangers of the latter path.
Finally, from my perspective anything that slows this buildout down is net positive over the long-term as the biggest risk and worst outcome is an *actual* overbuild. We are structurally short watts, wafers and increasingly capital. That is all good. God forbid we were still in ZIRP, TSM ramped capex to 3x their current plan and we weren’t short turbine blades: there would almost certainly be an overbuild and a bubble. That would be terrible. Slower and smoother for longer is better both for investors and for the world. A terrible cyber event is the most plausible near term bear case IMHO, but AI cyberdefense ramping quickly and I was happy to hear Sholto say that cyber probably becomes defense dominant in the near future.
Longer, threaded version follows:
Aaron Klein 🇺🇸 retweeted
A client check-in can take 35 seconds. Put it on the calendar.
@DerrickTKinney calls it “scheduled spontaneity.” You plan the call; your client gets an unexpected reminder that you’re thinking about them between meetings.
A few seconds on the phone can spare a client weeks of wondering whether you’re on top of things.
In this Beyond the Meeting conversation with @AaronKlein, Derrick shares how he made those calls part of his routine.
Watch the full episode.
go.contio.ai/derrick-x
Love seeing new models drop. Hate seeing new benchmarks drop. And it feels like every new model release has some way of acing the benchmark tests and that having zero relationship to the model's real-world performance.
Something has gotta give...
Blame the AI? They probably used Microsoft Word and Gmail too.
Here’s the bigger question: why are western news outlets so easily reprinting Iranian propaganda?
OpenAI revealed that Iranian operatives used ChatGPT to plant nearly 100 articles in Western news outlets under fake bylines; Separately, Anthropic disclosed AI agents submitting unauthorized visa applications and false police tips
ynetnews.com/tech-and-digita…
Aaron Klein 🇺🇸 retweeted
AI made the right way.
Giving people in real jobs tools to boost their impact in their work and more importantly amplify the impact of their relationships. 🚀
You have clients to win. Give Benny the renewal work.
Benny is Contio’s AI agent for benefits brokers. He reads the census, gathers rates and builds the proposal, drawing on your client emails, documents and conversations.
You decide which carriers belong in the mix and what to recommend. Benny does the work to put that strategy in front of your client.
That means a proposal to review instead of one more to build from scratch. And more room in your day for the relationships that grow your business.
Hire Benny for your next renewal.
go.contio.ai/benny-x
This is truly one of the most remarkable things I’ve ever seen. Michael and Susan Dell creating millions of new aspiring capitalists.
Bullish on America! 🇺🇸🇺🇸🇺🇸
"Today, we're excited to announce the $42 lunch."
Grok Bot is taking an interesting direction: choosing the best AI model for a task, even when another company built it.
Musk says it will use models including Claude, Midjourney and Suno alongside its own.
I'm torn about this approach. On the one hand, it's brilliant to ask for what you need without having to figure out which model is best at doing it.
On the other hand, I may not trust those other companies with my data — and SpaceXAI is making that decision for me.
How do you feel about this?
Aaron Klein 🇺🇸 retweeted
On Beyond the Meeting, @DerrickTKinney joined @AaronKlein to talk about a simple idea: advisors need coaching on their client meetings, too.
His coaching toolkit in Contio puts that into practice, with questions to guide the conversation and feedback on what you could do better next time.
Watch the full episode here: go.contio.ai/aiderrick-x
Aaron Klein 🇺🇸 retweeted
"If you want to make it rain, connect with pain."
"If you want a business that's booming, focus on the human."
"Make everything Derrick optional."
@DerrickTKinney talks the way he coaches. The least amount of words wins.
Behind each of those lines is a moment that nearly broke his firm. And a choice that transformed it instead.
Ep. 3 of Beyond the Meeting with @AaronKlein is out now.