Ron Itelman
@ronitelman.bsky.social
O'Reilly Author, "Unifying Business, Data, and Code" (2024), and Apress author, "The Language of Innovation" (2025)
Reposted by Ron Itelman
I think Claude Code has achieved AGI
November 22, 2025 at 9:09 PM
I think Claude Code has achieved AGI
GenAI for finance has serious flaws that without guardrails make it irresponsible for companies to adopt. Latest research I worked on showed ambiguity in a query for ONE word would swing answers by millions, and the AI model's ability to manage the risk was zero. Guardrails BEFORE hitting submit!
November 21, 2025 at 8:44 PM
GenAI for finance has serious flaws that without guardrails make it irresponsible for companies to adopt. Latest research I worked on showed ambiguity in a query for ONE word would swing answers by millions, and the AI model's ability to manage the risk was zero. Guardrails BEFORE hitting submit!
Reposted by Ron Itelman
For the first time ever, I had to make a payment in crypto.
I can now confidently say that crypto payments combine the speed of a dial up modem and the ease of updating the drivers on a 1998 HP Laserjet printer.
I can now confidently say that crypto payments combine the speed of a dial up modem and the ease of updating the drivers on a 1998 HP Laserjet printer.
November 18, 2025 at 5:58 AM
For the first time ever, I had to make a payment in crypto.
I can now confidently say that crypto payments combine the speed of a dial up modem and the ease of updating the drivers on a 1998 HP Laserjet printer.
I can now confidently say that crypto payments combine the speed of a dial up modem and the ease of updating the drivers on a 1998 HP Laserjet printer.
Here's how trustquery.com works for tabular data (like a .CSV):
1. As you type (or paste) your data into the chat, ambiguous columns are detected.
2. It notifies the user and asks a clarifying question
3. Any user feedback is added as metadata directly into the column name.
Open source (MIT).
1. As you type (or paste) your data into the chat, ambiguous columns are detected.
2. It notifies the user and asks a clarifying question
3. Any user feedback is added as metadata directly into the column name.
Open source (MIT).
November 18, 2025 at 8:56 PM
Here's how trustquery.com works for tabular data (like a .CSV):
1. As you type (or paste) your data into the chat, ambiguous columns are detected.
2. It notifies the user and asks a clarifying question
3. Any user feedback is added as metadata directly into the column name.
Open source (MIT).
1. As you type (or paste) your data into the chat, ambiguous columns are detected.
2. It notifies the user and asks a clarifying question
3. Any user feedback is added as metadata directly into the column name.
Open source (MIT).
Released open source library to turn any phrase into an interactive element: menus, api calls, guardrails, etc.
I'll be releasing more open source libraries on TrustQuery.com focused on pre-processing info for higher accuracy AI experiences.
Please reach out if you'd like to discuss!
Ron
I'll be releasing more open source libraries on TrustQuery.com focused on pre-processing info for higher accuracy AI experiences.
Please reach out if you'd like to discuss!
Ron
November 18, 2025 at 1:52 PM
Released open source library to turn any phrase into an interactive element: menus, api calls, guardrails, etc.
I'll be releasing more open source libraries on TrustQuery.com focused on pre-processing info for higher accuracy AI experiences.
Please reach out if you'd like to discuss!
Ron
I'll be releasing more open source libraries on TrustQuery.com focused on pre-processing info for higher accuracy AI experiences.
Please reach out if you'd like to discuss!
Ron
Reposted by Ron Itelman
ICYMI: Microsoft’s charge “implies a more than $12 billion quarterly loss at OpenAI, said Firoz Valliji, an analyst at Bernstein.”
That “would mark one of the largest single-quarter losses for a tech company in history.”
@jessefelder.bsky.social $MSFT
www.wsj.com/livecoverage...
That “would mark one of the largest single-quarter losses for a tech company in history.”
@jessefelder.bsky.social $MSFT
www.wsj.com/livecoverage...
November 7, 2025 at 7:17 PM
ICYMI: Microsoft’s charge “implies a more than $12 billion quarterly loss at OpenAI, said Firoz Valliji, an analyst at Bernstein.”
That “would mark one of the largest single-quarter losses for a tech company in history.”
@jessefelder.bsky.social $MSFT
www.wsj.com/livecoverage...
That “would mark one of the largest single-quarter losses for a tech company in history.”
@jessefelder.bsky.social $MSFT
www.wsj.com/livecoverage...
Reposted by Ron Itelman
Extreme Valuations for the S&P 500
Plotting the Warren Buffett indicator (US stock market cap to GDP) versus the Shiller cyclically adjusted P/E ratio shows that the S&P 500 today is at a historically extreme valuation.
Torsten at Apollo
Plotting the Warren Buffett indicator (US stock market cap to GDP) versus the Shiller cyclically adjusted P/E ratio shows that the S&P 500 today is at a historically extreme valuation.
Torsten at Apollo
November 6, 2025 at 12:20 PM
Extreme Valuations for the S&P 500
Plotting the Warren Buffett indicator (US stock market cap to GDP) versus the Shiller cyclically adjusted P/E ratio shows that the S&P 500 today is at a historically extreme valuation.
Torsten at Apollo
Plotting the Warren Buffett indicator (US stock market cap to GDP) versus the Shiller cyclically adjusted P/E ratio shows that the S&P 500 today is at a historically extreme valuation.
Torsten at Apollo
Reposted by Ron Itelman
This lack of reliability is I believe the real reason most companies haven't seen massive productivity gains with agents.
Just because a system *can* get it right, doesn't mean it *will*. And in LLM agent loops, errors earn compound interest.
Just because a system *can* get it right, doesn't mean it *will*. And in LLM agent loops, errors earn compound interest.
"Won't the AI just get smarter and figure it out?"
That's what a VC asked me when I explained why AI agents in healthcare are a safety disaster waiting to happen.
No. It won't.
Here's what they don't understand:
That's what a VC asked me when I explained why AI agents in healthcare are a safety disaster waiting to happen.
No. It won't.
Here's what they don't understand:
October 21, 2025 at 7:43 PM
This lack of reliability is I believe the real reason most companies haven't seen massive productivity gains with agents.
Just because a system *can* get it right, doesn't mean it *will*. And in LLM agent loops, errors earn compound interest.
Just because a system *can* get it right, doesn't mean it *will*. And in LLM agent loops, errors earn compound interest.
"Won't the AI just get smarter and figure it out?"
That's what a VC asked me when I explained why AI agents in healthcare are a safety disaster waiting to happen.
No. It won't.
Here's what they don't understand:
That's what a VC asked me when I explained why AI agents in healthcare are a safety disaster waiting to happen.
No. It won't.
Here's what they don't understand:
October 21, 2025 at 7:34 PM
"Won't the AI just get smarter and figure it out?"
That's what a VC asked me when I explained why AI agents in healthcare are a safety disaster waiting to happen.
No. It won't.
Here's what they don't understand:
That's what a VC asked me when I explained why AI agents in healthcare are a safety disaster waiting to happen.
No. It won't.
Here's what they don't understand:
This is not going to end well, and if your company's employees are putting valuable IP into their systems, what happens to the data if one of your competitors (or their IT suppliers/partners) buys a bankrupt OpenAI and gets a treasure trove of data?
September 10, 2025 at 4:39 PM
This is not going to end well, and if your company's employees are putting valuable IP into their systems, what happens to the data if one of your competitors (or their IT suppliers/partners) buys a bankrupt OpenAI and gets a treasure trove of data?
"You're absolutely right!" is fast becoming painful to read (if you use Claude Code, you know what I mean)
August 22, 2025 at 11:54 AM
"You're absolutely right!" is fast becoming painful to read (if you use Claude Code, you know what I mean)
If your business depends on A) an unprofitable business that B) needs to spend trillions of dollars, I have a question for you:
What happens to your business if your AI provider suddenly collapses?
What happens to your business if your AI provider suddenly collapses?
August 20, 2025 at 7:12 PM
If your business depends on A) an unprofitable business that B) needs to spend trillions of dollars, I have a question for you:
What happens to your business if your AI provider suddenly collapses?
What happens to your business if your AI provider suddenly collapses?
I have never sworn as much in my life as in the last 5 minutes when my super AGI literally can't comprehend the most basic instruction and build crazy complexity making mistakes no child would make.
August 20, 2025 at 6:34 PM
I have never sworn as much in my life as in the last 5 minutes when my super AGI literally can't comprehend the most basic instruction and build crazy complexity making mistakes no child would make.
Saw this excellent talk on OpenAI's intent modeling approach. Glad to see Intent Engineering and ambiguity detection and correction part of the AI conversation. This is exactly what TrustLoop is working on: aligning intent across business, data, and UX layers for natural-language-query to sql.
July 22, 2025 at 6:04 PM
Saw this excellent talk on OpenAI's intent modeling approach. Glad to see Intent Engineering and ambiguity detection and correction part of the AI conversation. This is exactly what TrustLoop is working on: aligning intent across business, data, and UX layers for natural-language-query to sql.
I've launched a new company, TrustLoop.AI, to eliminate the main source of hallucinations in AI, and my client said this phrase and it's only better than any marketing I could have come up with.
I'm not sure how much I'll owe him for it 😂
I'm not sure how much I'll owe him for it 😂
June 25, 2025 at 11:42 AM
I've launched a new company, TrustLoop.AI, to eliminate the main source of hallucinations in AI, and my client said this phrase and it's only better than any marketing I could have come up with.
I'm not sure how much I'll owe him for it 😂
I'm not sure how much I'll owe him for it 😂
Your AI is only as good as its understanding of your business context.
Smart leaders looking for quick wins: What's the smallest app tweak your team could make today to automatically capture more context about your workflows and business goals?
Smart leaders looking for quick wins: What's the smallest app tweak your team could make today to automatically capture more context about your workflows and business goals?
May 28, 2025 at 2:24 PM
Your AI is only as good as its understanding of your business context.
Smart leaders looking for quick wins: What's the smallest app tweak your team could make today to automatically capture more context about your workflows and business goals?
Smart leaders looking for quick wins: What's the smallest app tweak your team could make today to automatically capture more context about your workflows and business goals?
Working on private MCP Registries for companies working on AI & MCP, please reach out if you are available to chat, I'd love to hear about your core challenges.
May 27, 2025 at 6:10 PM
Working on private MCP Registries for companies working on AI & MCP, please reach out if you are available to chat, I'd love to hear about your core challenges.
"Sometimes magic is just someone spending more time on something than anyone else might reasonably expect."
- Teller
- Teller
May 18, 2025 at 3:39 PM
"Sometimes magic is just someone spending more time on something than anyone else might reasonably expect."
- Teller
- Teller
The biggest problem with "Vibe Coding" - identifying hacks so that you aren't leaving your systems vulnerable.
May 14, 2025 at 9:04 PM
The biggest problem with "Vibe Coding" - identifying hacks so that you aren't leaving your systems vulnerable.
Reposted by Ron Itelman
A bad sign, not just because of what it recalls historically, but for what it promises about the future.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
75% of US scientists who answered Nature poll consider leaving
More than 1,600 readers answered our poll; many said they were looking for jobs in Europe and Canada.
www.nature.com
April 28, 2025 at 11:56 PM
A bad sign, not just because of what it recalls historically, but for what it promises about the future.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
Ei incumbit probatio qui dicit, non qui negat
Translates to: "Proof lies on him who asserts, not on him who denies".
- Roman Lawyer Julius Paulus Prudentissimus 2nd century
Recently read about him on the topic of the history of due process.
Translates to: "Proof lies on him who asserts, not on him who denies".
- Roman Lawyer Julius Paulus Prudentissimus 2nd century
Recently read about him on the topic of the history of due process.
April 18, 2025 at 6:05 PM
Ei incumbit probatio qui dicit, non qui negat
Translates to: "Proof lies on him who asserts, not on him who denies".
- Roman Lawyer Julius Paulus Prudentissimus 2nd century
Recently read about him on the topic of the history of due process.
Translates to: "Proof lies on him who asserts, not on him who denies".
- Roman Lawyer Julius Paulus Prudentissimus 2nd century
Recently read about him on the topic of the history of due process.
Starting experimenting with the compounding error rate of agents that make small errors in small numbers of cells when "chatting with data" using LLMs, and the results are.... not good (if you care about accuracy).
April 11, 2025 at 8:10 PM
Starting experimenting with the compounding error rate of agents that make small errors in small numbers of cells when "chatting with data" using LLMs, and the results are.... not good (if you care about accuracy).
I'm imagining every European leader having discussions on a path to decouple any dependency on the US: reserve currency, supply chain, etc.
We have no understanding or experience of what that means, all economic models have no historical data for this, if and when it happens.
Unknown unknowns.
We have no understanding or experience of what that means, all economic models have no historical data for this, if and when it happens.
Unknown unknowns.
April 9, 2025 at 9:54 PM
I'm imagining every European leader having discussions on a path to decouple any dependency on the US: reserve currency, supply chain, etc.
We have no understanding or experience of what that means, all economic models have no historical data for this, if and when it happens.
Unknown unknowns.
We have no understanding or experience of what that means, all economic models have no historical data for this, if and when it happens.
Unknown unknowns.
"Capitalism is the astounding belief that the most wickedest of men will do the most wickedest of things for the greatest good of everyone"
- John Maynard Keynes, 1923
- John Maynard Keynes, 1923
April 7, 2025 at 1:53 PM
"Capitalism is the astounding belief that the most wickedest of men will do the most wickedest of things for the greatest good of everyone"
- John Maynard Keynes, 1923
- John Maynard Keynes, 1923
One of my favorite activities when working on innovation projects is what I call the "On Ramp." It's remarkably simple, yet I'm continually fascinated that clients almost never have one.
The act itself of creating an "On Ramp" forces concrete discussions about what is really important.
The act itself of creating an "On Ramp" forces concrete discussions about what is really important.
April 7, 2025 at 12:36 PM
One of my favorite activities when working on innovation projects is what I call the "On Ramp." It's remarkably simple, yet I'm continually fascinated that clients almost never have one.
The act itself of creating an "On Ramp" forces concrete discussions about what is really important.
The act itself of creating an "On Ramp" forces concrete discussions about what is really important.