Tom Lee (@TomMLee) header
Tom Lee (@TomMLee) profile picture

https://xitter.cf/t.co/G7H7B1IZow - An Entrepreneur with lots of bad ideas.

Bay Area
Joined January 2009
Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)Photo by Tom Lee (@TomMLee)
I haven’t lived here in awhile, but I see no other way to answer the hundreds of texts/DMs asking me if the world is, indeed, ending. In the span of six days, we’ve experienced; 11/12 — Microsoft outage 11/13 — Nationwide FAA ground stop 11/16 — Major US banking disruptions 11/17 — NSA’s classified internal network goes down We need far more analysis of why so much is breaking at once. I don’t want to get out over my skis here, and what I am hearing does not contradict the messaging that most of this is the result of screwy updates. But stepping back, the dots are starting to form a collective picture. We are seeing: • An explosion in severe and critical vulnerabilities, with AI accelerating discovery to 3X the number of vulnerabilities discovered in the same period last year. This is collectively being referred to as “vulnageddon” or “the vulnpocalypse.” • Mean time to exploit those vulnerabilities has collapsed to NEGATIVE SEVEN DAYS. AI allows attackers to find/exploit vulns far faster than software makers have time to fix them. • Powerful AI-enabled offensive tools like ARTEX have been thrown into the hands of every third tier nation state, cybercriminal and digital arsonist, which logically creates havoc. • This creates enormous pressure to compress patching windows and accelerate software updates, increasing the likelihood these fixes break things in unexpected, cascading ways. • Pressure to just “AI everything” means we are rolling out automated AI tooling on top of brittle infrastructure that can’t support it. My generation grew up in an extraordinary period of technological stability. We worked our way up the software stack, abstracting away layer after layer, until we lost a fundamental understanding of how the underlying systems actually work. We were told to “move fast and break things.” VCS told us to celebrate the idea that “software would eat the world.” Now we’re telling ourselves we can simply “roll out AI” across everything, as though nothing underneath will buckle. We never stopped to ask: SHOULD WE? Should we be digitizing every last nook and cranny of our companies, infrastructure, healthcare, homes, cars, bodies, lives? Especially when the complexity, interdependence, and fragility we were introducing overwhelmed our ability to understand/secure/maintain the systems we built? Nobody every asked: What if, in the name of progress, we were creating the ripest attack surface in human history? We are now entirely dependent on software systems that almost nobody understands end to end, that increasingly few people can repair, and that attackers can now probe at machine speed. We are hearing that incident response firms, the 911 for breach victims, are backlogged six months. Even those paid on retainer lack the bandwidth to triage every incident. The cavalry is disappearing. We are entering a period of unprecedented technological instability. “Software Beirut” is an incredibly apt analogy. We are in for a period of cascading failures, whether triggered by attackers, bad updates, or our own accumulated complexity. This, I’m sorry to say, is just the opening chapter.
715
4,006
650
14,557
2,431,890
Spanish Olympic skateboarder Danny León timed a mid-air trick to a solar eclipse, calling it "the move of his life."
126
1,567
115
12,318
614,252
Didn’t realize how huge the World Cup would be in America, until I saw 68,527 people fill Levi’s Stadium for a match between two countries with very little World Cup history (Austria and Jordan). That’s 300 seats short of a sellout for a Tuesday night match at 9pm local time.
258
1,036
65
32,456
3,571,768
Tried changing a flight through the @Delta app and online. Shows flights available, but when you click to change it gives you an error message. Eventually speak to customer service who then give you a different (much higher) price then you can see online. Why is this so hard?
1
1
103
Some epic thunder and lightening in Palo Alto right now
91
Tom Lee retweeted
Enterprise software is on the verge of another transformation. First came custom software: highly differentiated, built to fit. But expensive to maintain and hard to scale. Then SaaS took over: lower maintenance, continuous updates, greater efficiency. But also rigid cores, underused features, and roadmaps driven by vendors. We gained efficiency, but lost differentiation. Now we’re entering a third phase: AI-native services. AI fundamentally changes the economics of maintenance. When intelligent agents can generate, test, deploy, and continuously evolve code at a fraction of historical cost, the old trade-off breaks down. For the first time, companies can combine true customization with subscription economics. Software designed around your workflows and data, yet continuously maintained and improved at scale. This also changes what we mean by “services.” Services were never just manual work. They were human-scale work: revenue scaled with headcount, margins scaled with utilization. AI changes the lever. The next generation of services is human-led and AI-accelerated, architecturally repeatable, continuously evolving, and outcome-oriented rather than hour-based. At Globant, this shows up in AI Pods: AI-powered service units that design and continuously evolve living enterprise systems. Living software delivered as a service. There is no SaaS apocalypse. Standard platforms will remain massive businesses and critical pieces of enterprise infrastructure. But for many use cases, a smarter alternative now exists. Custom software gave us differentiation. SaaS gave us efficiency. AI-native services bring both together. The services industry isn’t going away. It’s being upgraded.
34
57
7
278
46,369
Tom Lee retweeted
people always ask why i care so much about culture… why i read about it, write about it, post about it. it’s very simple: it’s the only layer that actually matters. everything else… tech, markets, relationships, politics is just emergent behavior stacked on top. they’re lagging indicators of cultural mood. you can’t build, invest, lead, or even live coherently if you don’t feel what people want, fear, love, imitate, & aspire to right now. if you’re a ceo, vp, or vc, your only real job is reading people, zeitgeist, & timing. everything else is paperwork.
129
192
45
2,083
221,208
lol- I think we do more than that in a week at one customer. B2B SaaS is dead in 3 years.
Salesforce is about to pass 1M AI agent conversations on our own help.salesforce.com — working side-by-side with our 1M human agent conversations in Salesforce Service Cloud. AI Agents + Humans = Customer Success. ❤️🤖🤝 #Dreamforce #AI youtu.be/d-s-3V56o54?si=APcQ…
366
the united states is not at war with iran. it’s more of a special military operation.
539
579
164
5,174
525,970
As AI becomes more dominant, most people have a small window to make real money. You should take that window & run.
112
161
28
2,772
424,141
In the Fortune 500 average annual turnover from 1995–2011 was about 39 companies, versus 29 per year in 1955–1993, and in the late 2010s–early 2020s it has been on the lower end of that range (often under 5% of the list changing yearly). I suspect starting in early to mid 2030's I suspect it will explode.
14
7
92
19,742
When you start to use more AI Agents that do long running work in the background, it becomes clear that software is going to look very different in the future. Right now the vast majority of software was built to enable people to do all the work. Next, we saw a brief period where software has evolved for AI to be an occasional assistant in helping with that work. You can chat with AI in-line, it provides useful suggestions or data retrieval, and so on. But, as Agents get more powerful, this paradigm will likely flip. At least for some important chunk of software. As the AI Agents get more powerful, and you don’t need to go back and forth with them at every point, the software starts to become a tool for managing what AI Agents are doing for you. And software for people managing how agents work will look different from software that enables people to work. Software becomes about reviewing the AI Agents’ work, creating task queues, seeing the status of work as it’s happening, being able to interrupt the work when necessary, checking audit trails, managing parallel threads, connecting work outputs from multiple agents, and more. A large portion of software will likely evolve toward supporting this new approach. But a lot of software equally will need to be reinvented along the way. This is going to be both a crazy time for incumbents and a window where lots of new startup opportunities emerge.
51
111
20
820
246,425
Announcing @OpenAI Pioneers, a new program for ambitious companies building with our API. Selected teams will partner closely with us on domain-specific evals and custom fine-tuned models to advance AI product intelligence in their verticals. Apply to the first cohort below 👇
44
135
24
1,519
600,394
Enterprise software wasn’t built for humans. It was built to trap them. And now we’re trying to duct tape AI on top of it and act surprised when nothing changes. Every past platform shift was about how humans interacted with computers. Mainframe to desktop. Desktop to mobile. On-prem to cloud. The user was always the same: a person. This time, the user has changed. We’re building systems for machines, specifically for agents that can reason, decide, and act. That’s not a UI shift. It’s a total reset of how enterprise software should work. Legacy systems assume a human is always present to move a mouse, fix bad inputs, interpret context, and chase things down. Agents can’t do that. They need infrastructure built for them, structured workflows, constant feedback, observability, and self-evaluation. But the big software vendors aren’t going to lead this shift. They can’t. Their business depends on seat licenses, manual workflows, and keeping humans in the loop. The so-called “system of record” is a relic. If Excel disappeared tomorrow, half of enterprise software would stop working, and the other half would finally be exposed for the scam it is. Everyone knows this, but no one at the top wants to admit it. The system of record never truly existed. So what do the incumbents do? They slap some AI on top of their old tools. A chatbot in a dashboard. A sentence autocompleted in a text box. Cosmetic upgrades to avoid rewriting their own playbook. It’s not innovation, it’s fear of losing control. At Distyl, we’re not trying to retrofit the past. We’re building agent-first systems that can actually operate autonomously, systems that learn, improve, and replace brittle workflows instead of dressing them up. Most of the industry hasn’t caught up, but a few forward-looking enterprises have. They’re no longer asking if these systems work. They’re scaling them, optimizing them, and rethinking how their organizations operate at the most fundamental level. Enterprise software is undergoing a complete reset, and it will be a big shift for incumbents.
84
The Righteous Gemstones is Americas Father Ted
1,201
A royal podcast? Did not have that on my 2025 bingo card
1
179
There is an evals crisis; but I don’t think the problem is associated with benchmarking or even running evals. Companies need access to *people* who can develop evals by applying their domain expertise; and need systems to transform user feedback (regressions) into evals.
Replying to @TheZvi
My reaction is that there is an evaluation crisis. I don't really know what metrics to look at right now. MMLU was a good and useful for a few years but that's long over. SWE-Bench Verified (real, practical, verified problems) I really like and is great but itself too narrow. Chatbot Arena received so much focus (partly my fault?) that LLM labs have started to really overfit to it, via a combination of prompt mining (from API requests), private evals bombardment, and, worse, explicit use of rankings as training supervision. I think it's still ~ok and there's a lack of "better", but it feels on decline in signal. There's a number of private evals popping up, an ensemble of which might be one promising path forward. In absence of great comprehensive evals I tried to turn to vibe checks instead, but I now fear they are misleading and there is too much opportunity for confirmation bias, too low sample size, etc., it's just not great. TLDR my reaction is I don't really know how good these models are right now.
13
17
9
177
47,238
Tom Lee retweeted
🎄🎅starting tomorrow at 10 am pacific, we are doing 12 days of openai. each weekday, we will have a livestream with a launch or demo, some big ones and some stocking stuffers. we’ve got some great stuff to share, hope you enjoy! merry christmas.
1,258
1,986
916
23,348
2,346,556