Daniel Leon

One backend. Infinite client personalities. Use Polybot to run customizable WhatsApp AI assistants at scale. FastAPI, optional Lambda workers, Postgres context/session memory, provider adapters, Stripe access gating. github.com/leontramontini…

So much of getting better at anything is knowing where and when not to exert effort.

@BetterCallMedhi Sounds amazing. I'd love to hear how you'd pull it off with no connections no nothing there and a small budget. Asking because I am really considering it.

come build in shenzhen makerspaces open 24/7, every electronic component on earth within 30 minutes by metro, supply chain partners ready to to manufacture your prototype next week & VCs who actually understand hardware because half their partners are former engineers from huawei BYD, DJI or CATL who shipped real products before they wrote checks back home you pitch a SaaS deck to a 26 year old MBA who has never opened a soldering iron, here you sit with a guy who designed the camera module of the mate 60 pro & he asks you about the bill of materials in the first 5minutes the gap is widening every quarter and most of you are still arguing on twitter

Whenever someone starts his post with "should not be sharing this" I trust them and skip it, not because they shouldn't be selling this but because you fortunately end up building resistance to attention hijacking techniques.

Trying to really be trying 100% with my 100%, but it is very, very, easy to fool oneself.

@phosphenq very rusty in math but problem 3 might be solved even faster just using the binomial distribtuion 5!/4! * 1! = 5 (4-1=3)

@Dave_a_tron @wasphyxiation ahoy was so much better

The greeting "hello" did not exist until the telephone because there was no need for a context-free general-purpose greeting until phone calls

The criticism of @karpathy/@yacineMTB seems to invert the claim. "You can outsource thinking, not understanding" does not mean that outsourcing thinking gives you understanding. It means that an LLM can generate the reasoning, but it cannot make you understand it. So replying “but if you outsource your thinking, how could you understand?” is not a rebuttal. It is the point.

@FSfakiotakis @paulg Meta

@paulg What successful company or industry has actual contempt for its customers? In addition to porn and narcotics...

Don't join a company or industry that has contempt for its customers. You can make a lot of money that way, and of course it gives you a feeling of superiority, but you'll never do great work for a market you despise.

Good games make you forget everything outside the game, but also that they are game. Pretty similar to life.

For my first one and a half year of doing a “startup” , I took the do things that don’t scale too literally and ended up being a freelancer . When I realized, I didn’t even have a business but the worst job ev

I am a mindless mindlessness addict. I get my hit by consuming content.

Is there going to be a platform where only authenticated AI agents talk, trade and discover each other? If so , who is going to succeed building it?

In a time where everyone wants to be a founder, I am feeling more and more inclined to focus on becoming an independent researcher. Not because I have anything against academia or startups, but because both paths can distort my incentives. Academia can reward my need for approval, the security of being tied to an institution, and make me conflate citations and publications for what I would be supposed to do, which is truth seeking. Startups can reward my identity of being bold, and make me feel good when saying it. Sometimes I feel it has become the adult version of wanting to be a professional athlete when we were kids. But why do I really want to become one? Both things have a high risk of doing something we are (or at least I am for sure) very good at, which is deluding ourselves when the incentives are strong enough. If I really want prestige, it is easy to say to myself that I want to be useful. But do I really want that, to the bottom of my heart, or do I say it because it is the correct thing to say, because I have heard Elon and others say it? Independent research has a different incentive structure. If I have limited time, limited resources, and no remuneration for faking work or usefulness, I am pushed to choose something that actually provides value, no matter how marginal initially. Otherwise there is almost no reason to keep going. That becomes a first filter against self-deception. If I do not get past here, maybe I never really wanted what I initially told myself to begin with, which was to find new ways to create stuff that excited me and provided some kind of value. And today, starting is not as daunting as it used to be. I can download Codex, pay for one or two 20-dollar subscriptions a month, read papers, use open-source models, run benchmarks, and get up to date with a less resource-intensive research area. Of course, not everywhere. But in some undercrowded, software-adjacent areas maybe I can. I believe the gap between outsider and useful contributor is much lower than ever. The hard part is that the path feels indirect. At the beginning, it feels like I am moving in the wrong direction. I am not coding apps that no one will use (strange how productive this feels), not working for clients that I have to have, but that do not really scale up to something. I am reading, creating experiments, evaluating, benchmarking, noticing shortcomings. Who knows, maybe I am only fooling myself because I really like it. But maybe... maybe I find a real gap. Maybe I build something like a better retrieval engine for trial-to-patient matching. Maybe that becomes useful to a pharmaceutical company. Maybe it becomes a product. Maybe it just teaches me how to think properly about a field. I do not want to sound like an expert (in case I, in any way, do, I am definitely not). This is just a really heartfelt intuition. What I want to say, I guess, is that horizontal execution in software is losing value fast, and sales and distribution in social media are getting saturated, so I might as well choose a more interesting, and slightly less taken, path.

@incentivising Completely true, and also the reason some people say, correctly, I believe, that the workforce is populated with bureaucratic midwits signaling fake work

The most underused game theory move in professional environments is intentional underperformance in low-stakes situations. Because when you solve every problem immediately, you become the person who is expected to solve every problem immediately. Remember, the road to hell is paved with good intentions, and your ceiling eventually becomes your floor. When something is expected, the recognition disappears. But strategic underperformance in situations that do not matter enables you to preserve the impact of your performance in situations that genuinely matter. Professional environments are all about managing perceptions, and few learn that far too late.

@saen_dev Was excited for having real hate, then I realize you're just massively posting with AI :(

Recursion applied to technical debt. The ASCII animations are probably the most debuggable part of the whole thing.

I Vibecoded an app that explains your vibecoded code so it can explain itself. It's still buggy and it has a lot of work to do, what I'd save from it are the ASCII animations. Claude made the ship and Gemini the pirate. As usual, had a lot of ideas to continue, but life happens and time is limited, so now it is a cemetery of ideas and good intentions. I actually hope I'd never started, I was testing OpenClaw with Codex at the time via Telegram and it one shotted the first version, after testing it I couldn't stop playing with it for a while. I thought someone might like the idea and continue to play with it, so here's the repo in case you might like to. github.com/leontramontini…

I Vibecoded an app that explains your vibecoded code so it can explain itself. It's still buggy and it has a lot of work to do, what I'd save from it are the ASCII animations. Claude made the ship and Gemini the pirate. As usual, had a lot of ideas to continue, but life happens and time is limited, so now it is a cemetery of ideas and good intentions. I actually hope I'd never started, I was testing OpenClaw with Codex at the time via Telegram and it one shotted the first version, after testing it I couldn't stop playing with it for a while. I thought someone might like the idea and continue to play with it, so here's the repo in case you might like to. github.com/leontramontini…

@kamal_razzak @binghott @iamshackelford @DenneyDara @sourfraser @MatthewGattozzi @pkennedy93 @thedennis @harrydelmege_ @heyitsalexP strat

I interviewed all of the best creative strategists in the world. @binghott. @iamshackelford. @DenneyDara. @sourfraser. @MatthewGattozzi. @pkennedy93 @thedennis. @harrydelmege_. @heyitsalexP. (thank you so much guys, you're all the best) If you read this document you will be able to become, train, & hire the best advertiser/creative strategist in the world. I promise you that. Hiring and training creative strategists is one of the most expensive mistakes you can make if you get it wrong. So I asked the best in the world: what separates the ones who actually produce winners from everyone else. I wrote it all up in one doc. I put a lot of time into this and there literally 0 AI, just 14 pages of straight sauce. reply "STRAT" and I'll send it over.