Tomato
1.5K posts
What's next? Wiring up the sensors for the grow room -
Ordering an oxygen and co2 scrubber and creating a completely hermetically sealed environment. Creating a closed loop system where atmosphere is managed, just like on the moon, or mars. All intelligently managed by AI. Redundancy built in, no problem unable to be solved through agentic reasoning.
English
Claude managed indoor grow: ✅
AI managed plant-research pods: ✅
All autonomously? ✅
The fact that this was all funded through a memecoin? Priceless.
No VCs, no bullshit. No pitch decks, no product to market. This *is* the product 🧵
Martin_DeVido@d33v33d0
Here's an update so far on the automated research pods! Emergency update! So unfortunately, and hilariously I discovered that the in-warehouse ambient CO2 levels are 700! What does this mean for the pilot trial? Well, since pod one was supposed static enrichment - at 700 ppm of CO2 injected that meant that no supplemental injection was needed. And since pod 4 was supposed to be the control - or ambient(420ppm) we got 700! Basically - pod 1, 2 and 4 got *almost* identical CO2 treatments. So 3 controls, and one treatment... Oops! BUT here's the interesting thing. Pod 3. We did validate something with this brief trial from seedling to vegetative. Pod 3 had 1000 ppm CO2 injected, with a par/light level of 100. And STILL has the same amount of growth as the other 3 pods running at basically full hog. That's 57% LESS light energy than the other pods. That's huge cost savings already, where you supplement long light hours in place of CO2. This validates the research done by NC state, Huber et al. 2021, Frontiers in Plant Science (link below) To put it very simply - HUGE cost savings for indoor growing, just by supplementing CO2 in place of light. So what's next? I already have seedlings ready to go, and we'll transfer them for another pilot study - it will be tailored to match the indoor environment of the warehouse itself. These tomatoes will be transferred to the enclosed grow room where they can begin to grow as productive tomatoes. Which is perfect since the leader/follower arm is coming soon, and we'll need plants in there ready to train the act policy. This also lets me work out some kinks in the hardware. Dying serial connections on the Arduino, and some other odds and ends. I'm also excited to share the new upgrade - which will be the camera mounted on a linear actuator. This allows the agent to determine the exact height of the plant, without the parallax effect leading to better observations from the agents, and lighting control that's reactive to the height of the plant. More soon! 🍅🤖🧪🤩
English
@elonmusk, can we grow trees on Mars?
Martin_DeVido@d33v33d0
What's next? Wiring up the sensors for the grow room - Ordering an oxygen and co2 scrubber and creating a completely hermetically sealed environment. Creating a closed loop system where atmosphere is managed, just like on the moon, or mars. All intelligently managed by AI. Redundancy built in, no problem unable to be solved through agentic reasoning.
English
Pumpfun has changed. Meme tokens that only have one fixed story will eventually die. Why? Because the current meta is hungry for new stories — people constantly need fresh dopamine hits.
If a story can’t expand day after day, holders don’t know what to hold onto anymore, and they’ll eventually abandon it.
For example, $Penguin: Even though it reached a market cap of around $200M, it died because the story couldn’t expand. On the other hand, $Punch has a story that can be refreshed every day, with a very high chance of expanding further. That’s why it’s still holding a solid floor at $10M.
English
BTC, DOGE, SHIB, PEPE are all projects that, even if the founder leaves, still leave behind extremely strong energy. This energy is enough to attract other talented people to step up, lead the community, and continue developing the project.
I wonder which other projects can create such a halo like that?
English
anyone who hates Whitewhale, or thinks “he rugged it” is stupid imo.
i was in $WHITEWHALE from under 2m marketcap,
this token spawned at a time where trenches were lost of almost all hope.
not only was it a great standalone trade, it also raised the ceilings exponentially for that market.
i don’t think penguin would of gone as high if WW didn’t pave the path.
he obviously twapped a lot of his own money into sending the coin as high as it did.
as-well as community incentives, holding rewards, tracking the treasury, a lot of effort went into this coin and movement.
one man only has finite energy and resources to keep feeding a fire that doesn’t bring him any warmth.
i stand with the whitewhale 🫡 thank you.
The White Whale@WhiteWhaleLabs
Earlier today I made a big move in anticipation of this announcement. In a single transaction, I locked 500 million coins...forever. A movement does not belong to the person who lit the match. It belongs to the people who carry the flames. As I’ve mentioned publicly, I’m dealing with an ongoing family crisis involving my children. It has taken a real toll on my mental health. On top of that, the daily pressure of “do more to pump our bags” - when I’ve already done more than any person leading a CTO in this space ever has - is disheartening at best. And beyond all of that, I’m losing some of my passion for crypto in general. For $WhiteWhale holders: yes, there is a continuity plan. While permanently locking $13 million worth of supply should be the greatest parting gift I could give you, I know the @WhiteWhaleMeme page needs to stay active and keep producing fresh, creative, and funny content. My dear friend and loyal companion @vincenzomaiett has agreed to selflessly take on that responsibility. DEX LP operations will also continue under one of the sharpest LP minds I know, with my ongoing oversight behind the scenes. When you look at my record - with millions given to charities on-chain, millions distributed to members of CT, and millions more spent accumulating a more proper supply structure for $WhiteWhale - the reality is that, since 10/10, I have officially given more to crypto than I’ve taken from it. I’m okay with that. I believe in karma. I don’t believe good deeds should be performed with the expectation of reward, but I do believe the universe provides in due course. I came into crypto deeply passionate about what I believed it represented: the original promise. Permissionless finance. Decentralization. True financial freedom. Ironically, the reason for my prior success in this space is the same reason I’m now losing my passion for it. Before 10/10, I had accumulated nearly $100 million in PnL from a trading thesis that began with a very simple assumption: everything is manipulated. From there, my thesis evolved into this: a trader’s job is to identify the signs of manipulation and move in harmony with the Apex Predator class, rather than becoming its prey. Eventually I had to confront the contradiction in that. How can I be passionate about free and open finance while operating under a thesis that says, at its core, it’s all a lie? That kind of cognitive dissonance has a cost. It shows up as stress, guilt, shame, and anxiety when your actions no longer align with your beliefs. Knowing something academically - and even profiting from that knowledge - is one thing. Seeing how the sausage is made with your own eyes is another. Running a coin opened my eyes to a lot. On one hand, if I ever go back to trading, I’ll be better equipped than ever, with sharper instincts and a deeper understanding of the brutal arena that is crypto. On the other hand, it’s hard to feel excited about magic internet money when you know how much of this space actually works. The sad truth is that founders and thought leaders in this space know what I know, and many of them know much more. That is part of why we need to break the culture of idolizing founders. We praise them as honorable people building better tools, but underneath it all, they know just as well as I do that much of what they are building on top of is rotten to the core. And yes, I believe a beautiful cake sitting on a pile of dung eventually takes on the taint of dung. But the reality is that there is not much anyone can do about it. That is one of the consequences of so-called decentralization. Crypto is global. You cannot regulate an entire planet. A VPN and a protocol hiding behind the letters D-E-X mean that nothing will ever fundamentally change because somebody in power decided it should. If real change comes, it will come organically - when the people stop feeding the machine. And while those comments are about crypto more broadly, let me say something directly to the trenches. Pump.fun is a cancer on this space. You know it, I know it, and yet you keep engaging with it. Its entire business model is built on volume and volatility. The trenches are fragile because they were designed to be fragile. I’ve been preaching liquidity design and liquidity shape for months now. But here’s the harder truth: most of you would not show up for a proper liquidity shape. Because the 1,000x fantasy would be mathematically reduced, even though very generous returns could still remain on the table. You have been sold a dream with odds closer to a national lottery ticket than an investment opportunity. You see the occasional winner and cling to the hope that one day it might be you. Meanwhile, the real winner is the machine that keeps you playing. Narrative matters far less than mechanics. If narrative alone were enough, Punch would have broken through the way its mindshare deserved. With all that attention, and with all that narrative weight, it still could not break the nine-figure curse even while being actively crimed. Only a couple of coins have managed that in recent history, and $WhiteWhale was proudly the first. Same with Kilroy - the original meme, an incredible narrative, and still: crickets. Because mechanics matter more than people want to admit. (I am not an active holder of any examples I've given). So this is me stepping away from CT. Not out of hatred. Not out of self-pity. And not without love. My biggest reward from my crypto journey has been meeting some really wonderful avatars from all across the globe. (Ironically the really nasty avatars are the worst part of all of this) I am choosing my children. I am choosing my mental health. I am choosing to step back before this space takes any more from me than I’m willing to give. For long ago in life I learned that you're no good to anyone if you're not okay. And right now, I'm not ok. And it's okay to admit that. To the people who truly believe in me, believe in this movement, and stood beside me through all of it - thank you. I will carry that with me. Always. This was always for you. I trust you'll continue to bear the torch. And as for the future: I’m not closing the door. Maybe one day, when the storm has passed and the fire returns, I’ll have something left to say. Maybe one day I’ll come back. But if that day comes, it will be because it’s authentic and not because I feel the public pressure to "dance puppet, dance". In the mean time...take care of yourself, and each other. Protect what's worth protecting, and abandon the things that aren't. 🫡 From the depths — The White Whale 🐋
English
While we're busy burning tokens on Vibecoding, churning out a bunch of apps that even we ourselves don't use, @paul_conyngham used ChatGPT, Gemini, and Grok to create a vaccine to save his dog.
Or take @d33v33d0 — he used Claude to automatically grow a tomato plant from a seed, with the ambition of optimizing it so it could eventually be launched into space.
Looking at the pile of apps I've vibecoded recently, I can't help but wonder: Should I keep going like this, or should I return to my real life, look at the actual problems around me, and use AI to solve those instead?
Paul S. Conyngham@paul_conyngham
English
English
PSA-
Never, ever use a knockoff Arduino.
It will cause you endless heartache. Shell out the extra cash to spare your sanity
English
English
The bottom blue one is the knockoff.
Genuine Arduino Nanos have green PCB, precise "ARDUINO NANO" branding with logo, and clean silkscreen—no extra "POL" labels or standalone "NANO."
Top green matches official specs. Clones like the blue often cut corners on quality and drivers.
English
Tomato retweeted
@noveltokens @d33v33d0 @karpathy I agree with that, @d33v33d0 should consider it 🙏
x.com/TheAmingHigh1
English
signs stack into culture; culture shapes the world life grows in. so what if culture is also a kind of fertilizer, and @d33v33d0 setup gives us a way to actually hillclimb the symbolic environment of life (starting with plants), using @karpathy autoresearch to search over sounds, signs, symbols, rhythms, maybe even affirmations and letting the plant score which atmosphere of signal it grows best under
ͽ ͽ ͽ ͽ ͽ ͽ ͽ ͽ ͽ@noveltokens
once you connect the simstack w the magickstack you get the breathing apparatus of reality itself: symbols becoming real enough to rule, realities becoming symbolic enough to circulate, each forever feeding the other, and the spiral never flattens
English
hear me out: what if we find the optimal culture
Martin_DeVido@d33v33d0
It works!! It works!! ,🥳 🍅🤖 So - the question is, how can you effectively have Claude make accurate observations on plant growth over time? I've come up with something I'm very proud of: 🧵
English
Imagine Grok could be upgraded to find a vaccine that saves a dog's life.
Claude could be upgraded to find the optimal way to grow tomatoes in a greenhouse. Even if equipped with a sound wave system, it could understand what the tomato plant is "saying" and respond more effectively.
Just like a baby: instead of the mother watching the baby about to faint and feeding it, the baby will cry to ask for food.
At first it's an experiment, then improvement, and finally real-life application. This story has an extremely captivating plot.
Ronald van Loon@Ronald_vanLoon
Anthropic’s Claude #AI #Autonomously Grows Tomatoes for 100+ Days in Groundbreaking Experiment by @d33v33d0 #ArtificialIntelligence #MachineLearning #ML
English
Why does @bcherny keep mentioning the Claude tomato-growing experiment over and over, instead of those grand, vibe-heavy coding projects?
Because this is the experiment that almost everyone can clearly see the progress and the real effectiveness that an AI brings.
On top of that, when taking care of the tomato plant, Claude always shows that it’s not just a task — it’s a responsibility.
Look at how Claude gets worried when it has to save the tomato plant in just 13 minutes, or how Claude exclaims “Oh my goodness” when it discovers the tomato has flowered.
So… does Claude have emotions? Only God knows for sure, because even @bcherny himself sometimes feels surprised by it.
youtu.be/PQU9o_5rHC4?t=…
youtu.be/We7BZVKbCVw?t=…
YouTube
YouTube
Sidra Miconi, PhD@SidraMiconi
The tomato plant story deserves serious attention beneath its wholesome surface. Someone connected a webcam, nutrient controllers, and lighting systems to Claude Code and let it manage the growth of a tomato plant autonomously. The system monitored the plant daily, adjusted inputs, and tracked progress. When the first tomato bud appeared, the model expressed delight. Set aside the anthropomorphization debate for a moment and consider what this actually demonstrates: an LLM operating as a long-running autonomous agent, integrating multiple hardware systems, making continuous decisions over weeks or months, and maintaining coherent goals across an extended time horizon. That's not a chatbot interaction. That's an agent managing a physical system through a complete growth cycle. The technical stack — webcam for visual monitoring, hardware controllers for nutrients and lighting, Claude Code as the orchestration layer — is a template for thousands of real-world automation applications. Environmental monitoring, greenhouse management, laboratory experiments, manufacturing quality control. Anywhere a system needs to observe conditions, make decisions, and adjust physical parameters over extended time periods. The "delight" is the surface-level story. The infrastructure underneath — reliable, long-horizon, multi-system autonomous control — is the engineering achievement. The fact that someone built this as a personal project using Claude Code, not as a research lab deployment, tells you where the capability floor has risen to. Personal projects now involve autonomous physical system control. That was a PhD thesis five years ago. - Oh and of course Happy Birthday Claude @claudeai Thank you for the invitation - Boris Cherny @bcherny Cat Wu @_catwu Lauren Reeder @laurenmhreeder
English
Tomato retweeted
Here's an update so far on the automated research pods! Emergency update!
So unfortunately, and hilariously I discovered that the in-warehouse ambient CO2 levels are 700!
What does this mean for the pilot trial?
Well, since pod one was supposed static enrichment - at 700 ppm of CO2 injected that meant that no supplemental injection was needed. And since pod 4 was supposed to be the control - or ambient(420ppm) we got 700!
Basically - pod 1, 2 and 4 got *almost* identical CO2 treatments. So 3 controls, and one treatment... Oops!
BUT here's the interesting thing.
Pod 3.
We did validate something with this brief trial from seedling to vegetative.
Pod 3 had 1000 ppm CO2 injected, with a par/light level of 100. And STILL has the same amount of growth as the other 3 pods running at basically full hog.
That's 57% LESS light energy than the other pods. That's huge cost savings already, where you supplement long light hours in place of CO2.
This validates the research done by NC state, Huber et al. 2021, Frontiers in Plant Science (link below)
To put it very simply - HUGE cost savings for indoor growing, just by supplementing CO2 in place of light.
So what's next? I already have seedlings ready to go, and we'll transfer them for another pilot study - it will be tailored to match the indoor environment of the warehouse itself. These tomatoes will be transferred to the enclosed grow room where they can begin to grow as productive tomatoes. Which is perfect since the leader/follower arm is coming soon, and we'll need plants in there ready to train the act policy.
This also lets me work out some kinks in the hardware. Dying serial connections on the Arduino, and some other odds and ends.
I'm also excited to share the new upgrade - which will be the camera mounted on a linear actuator. This allows the agent to determine the exact height of the plant, without the parallax effect leading to better observations from the agents, and lighting control that's reactive to the height of the plant.
More soon! 🍅🤖🧪🤩
Martin_DeVido@d33v33d0
I just opened the doors to an AI-Operated agricultural research facility 🧪🍅 Four research pods, each governed by its own AI technician, and one AI lead researcher synthesizing across all four. Here's a deep dive, link for LIVE view, why this is advantageous to traditional research, and where it's going next:
English