Luis Lascurain (@luislascurain) header
Luis Lascurain (@luislascurain) profile picture

Reducing the cost and risk of investing in agriculture | Chemical Engineer | Stanford MBA

Joined September 2010
Photo by Luis Lascurain (@luislascurain)Photo by Luis Lascurain (@luislascurain)Photo by Luis Lascurain (@luislascurain)Photo by Luis Lascurain (@luislascurain)Photo by Luis Lascurain (@luislascurain)Photo by Luis Lascurain (@luislascurain)Photo by Luis Lascurain (@luislascurain)Photo by Luis Lascurain (@luislascurain)
Did knowing math just become a lot more relevant? We’ll see those who understand these concepts build amazing solutions that were out of reach before. Excel made understanding finance more valuable, but less fun/valuable to add and subtract teapaw.org/openai/status/21075967…
We’re releasing a broad range of new mathematical results produced by an internal frontier model. We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results. github.com/openai/math
1
1
88
If people are not going to lose their jobs to AI, but to someone using AI… companies will face the same. Thus, should startups sell AI to companies or do the job and win by using AI?
12
3
102
145,287
So where does it leave it for startups - or many of them? Likely in the domain expertise services, where most can be done by an LLM but requires some “last mile” earned secret to finish the job, and in industries where incumbents will have a hard time adopting AI
1
2
112
Large LLM companies will try to be the “everything services” and you could say they allow 3rd party sales via the API. If you are in that bucket you risk facing an “Big LLM Basics” competitor
1
144
The clearest analogy to the past is ecommerce. Where “the margin was the opportunity.” The difference now is that it applies to services instead of retail
1
247
It likely depends on the job to be done, and the AI capabilities of the incumbemt vs the startup. There are going to be a lot of new amazing businesses related to AI - today they are mostly in the model and compute - but also many that will just do the job better because of AI
1
337
Bold move, and likely has a place at the legacy enterprise. Still, it feels like the time Yahoo and The NYT gave Google a search bar in their webpages. The system of record matters, but being the best at converting energy to outcomes is were the $$$ is
Salesforce $CRM and Anthropic just released this new commercial together
88
A side benefit of the data center build out is it will lower the costs for greenhouses that collocate (cheap to keep warm). Significantly lowering the cost of food. The next bottle neck for this is labor/robotics to operate those greenhouses
1
1
76
A clear example of precision being more valuable than complexity + economies of scale
It's crazy that we can make a V8 engine block for $10,000 but a PT-6 is well over a million.
116
Luis Lascurain retweeted
super excited to share that we were accepted into @StartX fall batch 2026. will be sharing more in the coming weeks about what we’re building at Rizal. onwards!
1
9
370
If you care about outcomes you care about evals. Similar as having OKRs or NPS - agents require evals. This is why we are constantly looking at how accurate, how fast, and at what cost our agents are performing
AI diffusion is far more rate limited by having good evals than most realize. The kind of evals that you see for every model release are incredibly helpful, but only tell you the shape of general AI progress and the relative capability level of models. The far bigger space over time are evals on all the major workflows that enterprises do, down to the specifics of an individual company. This will be a huge space over time because you can’t automate what you can’t assess the progress on. Enterprises will not be able to go just on vibes.
1
1
2
102
Every tech/services company will need its token factory. Selling tokens at a premium will require deep domain expertise, and likely trust. The less tokens they need for an outcome the higher the margin
1
34
🇲🇽🇲🇽🇲🇽
29
Luis Lascurain retweeted
No hay duda: AMLO ha sido el peor presidente de México. Corrupto, mentiroso, manipulador, metido con el narco (narco presidente), nefasto, mala persona, enfermo física y mentalmente, cometió crímenes de lesa humanidad contra México … vienen tantos adjetivos que se convierten en una idea: Jamás debió de haber llegado a la presidencia de México. Estábamos mal - en muchos temas - pero había instituciones, democracia, organismos intocables como el INE, el Poder Judicial, la Educación, la Seguridad … pero este inútil desmadró cualquier avance para convertirlo en retroceso. ¡Fue peor que Echeverría y eso es mucho decir! Puede ser que esto necesite más de 50 años para encontrar - más o menos - un cauce institucional, democrático … #AMLOelpeorpresidentedeMéxico FF
945
4,638
103
13,078
140,492
Luis Lascurain retweeted
Society doesn't reward truth. It rewards mastery of communication. Like it or not, that's the game.
4
19
108
4,871
Stop giving employees stock options. Give them @Polymarket/@Kalshi bets on company earnings / kpis. Gives more agency and causality to employees and removes macro/wall street distractions for employees
1
1
172
Luis Lascurain retweeted
219
1,649
731
12,419
13,278,609
I dont get why Word is not Cursor for lawyers. It already has “redline” and “comments.” Just add an agent sidebar
3
60
In a world where intelligence becomes a commodity, will price to book values become more relevant?
1
106
If replacement costs go down fast. One of your highest costs is the cost of capital. AI is supercharging this effect for software
1
142