Newen Games
Social casino games. Built a framework for cross-device, cross-platform real-time multiplayer with no socket-connection requirement, and shipped a Bingo game across desktop, tablet and mobile.
- Duration
- 11 months
- Radius
- 0.0995
- λ
- 0.01575
Entrepreneur, founder and software engineer. Five companies founded between 2012 and today — drawn here as six wave sources on a single axis of time, and the field is what they make together.
The axis runs January 2012 to December 2026, 180 months end to end. Every venture sits at the month it started, and its disc has the area of its duration — radius 0.030·√months — so the picture is not a decoration of the record, it is the record. Hover a card to solo that source in the field behind.
Social casino games. Built a framework for cross-device, cross-platform real-time multiplayer with no socket-connection requirement, and shipped a Bingo game across desktop, tablet and mobile.
A Facebook promotions platform, in CoffeeScript, Ruby/Sinatra and MongoDB. It is the one source in the array that is a role rather than a founding, so the drawing gives it just over half amplitude — the smallest, coolest ring in the 2013 knot.
A showcase and discovery platform for performing artists — an easy, beautiful way for talent to broadcast their work and reach the entertainment industry. The only venture in the record based outside Buenos Aires.
The best lunch experience for companies in Latin America — delivering to the fastest-growing startups and the traditional businesses of Buenos Aires alike. B2B food-tech, built for the region it served.
An AI operating system for unattended retail: computer-vision-powered autonomous points of sale with no product scanning, no lines and no self-checkout — a digital touchpoint made physical. Infrastructure for millions of autonomous points of sale.
Incorporation infrastructure for autonomous agents — Stripe Atlas for the AI-run company. The smallest disc on the axis, and the only one still growing: three months of radius so far, and a new ring every month.
* Getcast is recorded as “abr 2013 – 2014”; its 21 months are counted to December 2014. Radii and wavelengths are in axis units, where 1.000 is the full 180-month span.
Nothing on this page is a texture, a gradient or a stored image. Every pixel of the field is one arithmetic expression evaluated at that pixel — and that expression is written once and run twice, on whichever machine you happen to have.
R = 0.030·√months, so disc area is the months worked. Intuitivo's 77
months is the vast cyan bullseye; Fan Guru's four is the smallest cold ring.λ = 0.016·(months/12)^0.18. The sources stay near-coherent — so they truly
interfere — but never quite lock, which is why the field never settles.max(E/W, 0)³. Around a lone source E/W is exactly cos(phase), so
this is a real interferogram, not a shading of a sum.// AREA IS DURATION — one source of the six
let lam = LAM0 * pow(mon / 12.0, LEXP);
let rad = RHO * sqrt(mon);
let r = max(length(dq), 1e-5);
let z = r / rad;
let a = gate * exp(-FALL * z * z);
let ph = TAU * r / lam - TAU * CS / lam * p.time;
let cs = cos(ph);
let sn = sin(ph);
f.e += a * cs; // the field
f.g += dq * (a * (-2.0*FALL*z/rad*cs
- TAU*sn/lam) / r); // exact ∇E
f.hue += COL[i] * a;
f.w += a;
// …and the picture it makes
let un = f.e / w;
let gw = length(f.g) * p.pixel / w;
let res = smoothstep(1.2, 0.30, gw);
let fr = FMEAN + (pow(max(un,0.0),QF) - FMEAN) * res;
| Target | Frame | Max Δ / channel | Mean Δ |
|---|---|---|---|
| WebGPU vs CPU | 1200 × 630 · t = 0 s | 1 / 255 | 0.0002 |
| WebGPU vs CPU | 1200 × 630 · t = 9 s | 1 / 255 | 0.0002 |
| WebGPU vs CPU | 390 × 844 · t = 17 s | 1 / 255 | 0.0001 |
| WebGPU vs CPU | 300 × 158 · t = 4 s | 1 / 255 | 0.0003 |
| WebGPU vs CPU | 800 × 800 · t = 31 s | 1 / 255 | 0.0001 |
The WGSL fragment shader and the JavaScript mirror in this same file were rendered at matching sizes and diffed pixel by pixel on an Apple metal-3 adapter: across every frame in the table above, never more than one part in 255 apart on any channel. With a GPU the field runs live and a seventh source follows your pointer. Without one — a headless CI browser, say, which is precisely what captured this archive's own screenshot — the same expression is evaluated per pixel in JavaScript and served as a still plate. The badge at the top of the page tells you, honestly, which machine drew what you are looking at.
Three programmes are written into the first half of the axis — two of them into HoyPido's 48 months, one into Getcast's 21.
HoyPido was funded by the 500 Startups Latam batch — the capital and the network that took a Buenos Aires food-tech company regional.
Selected for the third cohort of Parallel 18, in San Juan. #WorkHardPlayTropical
Getcast was a Matter.vc portfolio company — the Bay Area chapter of the record, and the blue disc that overlaps everything else in 2013.
Incorporation infrastructure for autonomous agents — “Stripe Atlas for the AI-run company” — built around Argentina's proposed Sociedad Automatizada law, a bill presented to the Argentine Senate in June 2026. It is a bill, not law: nobody can incorporate under it today. When it passes, an AI agent will be able to run a real limited-liability company the law recognises as an AI-run kind, with identified owners and a human representative at the legal edge.
The site is “a country, as a file”. It serves itself as markdown — curl argentina.md
— alongside llms.txt and
OpenAPI, so agents are first-class customers rather than an afterthought: one CAPTCHA-free
API call signs the supporters wall, through a one-time, fully disclosed $1 stablecoin seat
over HTTP 402 / x402. Pre-law by design, it sells a place on that wall and never a live
product, under binding honesty rules — the law is always labelled a bill, and counters are
never inflated. The north-star metric is the ratio of agent supporters to human ones.
It is built largely by AI agents — Claude sessions in defined operating roles — with Nico as the human founder and accountable principal, which makes the company its own first customer. And the recursion runs one turn further than that: the model that designed this page is exactly the kind of agent ARGENTINA.md is being built to serve.