I used to only engineer software; then I learned you can engineer everything. Building AI agents to automate supply chains. Cofounder @ Elm AI | @Cornell CS
NYC/London
Joined June 2012
- Tweets1.4K
- Following602
- Followers498
- Likes488
either lovable is grossly overvalued or wix is grossly undervalued
wix owns base44 that blasted past 100m earlier this year, and is closer to 250m now.
wix has a market cap of <3b
1
131
we were a 4 person company when we signed our first public company
(we now have many publicly listed customers and are now an 11 people company)
PS: we are hiring a founding gtm/strat in NYC
How does a 4-person startup sign on a public company as a customer?
By making them billions of dollars... or in the case of one customer of NYC startup @sequen_ai, tens of billions.
___
"There's real impact in going into a business and having the ability to show that you were able to 3X or 4X a publicly traded company's annual revenue targets in just a couple of weeks," CEO Zoë Weil says.
"You're going in, and a company has 20 people working on this problem, and they haven't been able to have a lift in 18 months, and you come in a couple of weeks, you absolutely crush it."
For Sequen, CTOs have now proven the most valuable source of new business.
"One CTO, of a publicly traded company, said to another CTO: 'I know this is the most important decision you'll make as a CTO because this is the most important decision I've made as a CTO.'"
___
This episode is sponsored by @Atlassian and @chargebee.
Thanks to @NYSE for the studio!
we are still hiring for this role -- we have yet to find the right person to lead growth and strategy at elm ai in nyc!
supply chain executives are new to ai; they are finally curious and embracing it -- and with almost all our customers we are the first case study of useful vertical agents to manage their suppliers and factories
being our first growth role, this person would help educate execs, experiment with channels and overall build a truly ai forward gtm motion to 10x our current revenue.
if we do our job right, our technology will help make the production of physical goods higher quality, more affordable and more ethically sourced.
DM me if this sounds like you
im hiring our founding gtm lead!
probably one of the highest impact roles in the company right now
supply chain ai today, is what coding ai was in 2023 -- super early, barely any adoption, healthy skepticism.
we know how ai coding turned out, and it's obv supply chains are an *even better* application.
we are already working with some of the worlds most complex supply chains proving that this works incredibly well, and we want to scale fast!
work at the most impactful vector for ai diffusion.
(apply and dm me if interested)
this was me roughly 5 years ago
I really enjoyed programming, and the idea of "elegant solutions"
ai has changed so much, I can't ever imagine putting in this kind of thought in programming
but building software, organisations, societies is abstraction layered on abstractions
its best we learn to move up the abstraction chain with tact and grace
1
6
233
eh, people are getting too obsessed with "software factory" as a phrase. My best guess is that "factory" is seen as a derogatory thing.
Everything that is not art, should be produced en mass, in a factory.
As a heuristic, I like to think of art as anything who's primary objective is uncorrelated to increasing gdp, or improving peoples' economic or biological wellbeing.
We need big beautiful software factories. And data center factories. And vegetable factories. Drug factories. Math factories.
If something can be made in a factory, it doesn't mean you can't do it for art. I code for fun. I grow basil and mint for fun. Mathematicians can solve their toy problems for fun like I create toy databases for fun.
It's not that I disagree with @poteto, I think throwing shade at a software factory is not productive. She runs a factory. Anyone who merges 5000+ PRs is running a factory.
Her statement is akin to "You don’t need a car factory, you just need good engineering, well-defined processes, and rigorous quality control" -- that is literally a factory.
We should not be so precious about the word "factory".
why are so many companies building these solutions in house?
amp and devin are so good and keep launching new features -- how can an internal product keep up? who will keep updating things?
eg: devin came out with security, amp has a great app and orbs are so good
now that @conductor_build got acquired by @Meta, where's everyone switching to?
we've been building our own internally, called Central, but never with the intention of releasing it publicly
debating if we should launch it
til SAC Capital got caught doing insider trading on an alzheimer's cure, and were fined 1.8b after pleading guilty
the til part is, SAC Capital is now Point72, they rebranded to get away from the reputation SAC had after they got caught.
118
*with caveats
This is good advice. Actually, its great advice.
I didn't know it this when I first raised (and I feel there is wishful thinking from first time founders that investors will help)
but the white pill, if you will, is that you can do anything you want, it's your company, and you are the only person in the world who has the ability and responsibility to take the company to the commanding heights that it should be at
(and yes, i am happy with our investors!)
Generally speaking, the best investors are the ones who get out of the way the fastest.
"Value add" investing is mostly a joke. If your investors knew how to run your company, they'd be running a competitor.
1
244
i wonder what are the architectural and engineering challenges to make data centers monuments of human achievement -- probably one of the most impactful endeavors of our time
the rebrand is one good step, but we need to figure out how to kill the noise, make it beautiful, and give it the same aura as the great architectural feats of human civilization
openai might rebrand data centers as "ai compute campuses" as part of a public relations push after the midterm elections, according to axios reporting cited by techcrunch.
source: techcrunch
1
4
172
I agree it's clever, but it's also dangerous in that data centers are good!
it's doubling down on misinformation that will deprive more people of access to ai -- can a local model as good as even terra run on devices that average people have? not today.
(misinformation about data stealing, which small models also do, and water/energy usage, which local devices will also need)
can local models run 24/7 to run work? yes, only if your device is always on and connected to all the services it needs to be connected to (email, etc)
can your average person manage that? I doubt it
we need data centers to serve ai efficiently at scale to billions of people
a good analogy here is stating:
farming vegetables at home is really good but farming at scale is bad
Very clever positioning.
It basically counters every argument people have against AI and just gives you local intelligence -which no one should be opposed to.
Kudos on the launch @0xSigil
1
144
This sort of post hoc analysis is fun, but in practice has little to no alpha, and full of spurious correlations
In fact, it might even bias you away from actual alpha, and is against the core ethos of silicon valley historically
But it's fun to see, in a guilty pleasure way.
I redid the startup founder signals that lead to top tier VCs investing in the seed / A, after incorporating some feedback.
This added ex-frontier lab and cyber security near the top, IIT as a positive, and educated in China as a negative.
The main changes were;
- extended founding year to 2019-2026, ~2x'd the number of startups
- unified and added some signals
Here are the new signals:
3
322
I see a few comments and qt's about the network effect I am talking about with "personal" agents
what if muse had a muse network that allowed your muse to talk to your wife's muse? at that point does it make sense to be on poke or instinct?
it's going to be twitter vs atproto all over again
I am hoping that everyone uses a common agent communication protocol, but similar to imessage vs RCS, there will be special "in network" features for certain agent networks.
happened 1 month after $7.5M seed announcement
gg to founder for having the intellectual honesty and coming to the tough conclusion so fast, opportunity cost for being directionally wrong is very high right now
competing against muse, poke, instinct, and the half dozen others is incredibly hard, since it feels like network effects will lead to winner takes most (social media style)
