darkl0w

rrpp / @fcbarcelona

[darkl0w @darkl0w]

Match Day 🏆

#FCBarcelona #smsports #MatchDay

[darkl0w @darkl0w]

@aiscwork esto es cierto??!

[darkl0w @darkl0w]

@robrtcode que buena patron

[Chesny @chesny]

This startup built an AI that found Bluetooth vulnerabilities nobody knew existed.

Then the same system led the development of a rocket that lands itself.

Not two different models. One.

Autopoiesis is training AI on scientific judgment, the part where you pick what's worth chasing before you have any proof it works. Turns out that transfers across fields.

Still early. But if it holds, science

[darkl0w @darkl0w]

@Sonilo_music @chesny this is crazy wtf

[darkl0w @darkl0w]

@aiscwork real?

[darkl0w @darkl0w]

@chesny muy top

[Chesny @chesny]

los que llevan meses perdiendo tiempo cada vez que Anthropic recorta un modelo, sin saber que el manual de comportamiento se puede extraer y reciclar

[darkl0w @darkl0w]

@robrtcode guardado!!

[robert. @robrtcode]

ASÍ SE MONTA UNA SKILL EN DRIVEN

→ Partes de un método de research que ya usas. Una matriz de valoración, un screener, un desglose de earnings, lo que ejecutes cada semana

→ Se lo describes a tu Agente en una conversación normal. Sin código, sin configuración, sin conectar fuentes de datos

→ Driven convierte ese método en una Skill: una capacidad repetible construida sobre un framework concreto y citable

→ La Skill se ejecuta como código real, no como una respuesta de texto. Misma estructura y misma lógica siempre

→ Tira de más de 260 endpoints de datos profesionales en más de 30 categorías. Cotizaciones, filings, holdings, actividad de insiders, flujos de capital

→ Las Skills se revisan y se versionan, así que lo que construyes queda auditable

→ Desde ahí la publicas en el Skill Store, o ejecutas las que ya han creado otros inversores

→ Tu Agente lanza la Skill correcta por su cuenta según lo que le preguntes

Esa es la diferencia entre un prompt guardado y un proceso que funciona.

El Skill Store ya está live: https://t.co/4BZQEhQbm0

Si tu también quieres empezar a hacer este tipo de Skills y ahorrarte tiempo de vida, regístrate y dispondrás de un mes gratuito con la suscripción Pro

Guárdalo, ya me lo agradecerás

@getDrivenAI

[Chesny @chesny]

Everyone keeps comparing this AI wave to the internet boom, but there's one huge chunk of the economy the internet boom never actually touched.

Industrial B2B. The stuff that gets built, not clicked. Still runs almost entirely on relationships and phone numbers.

Gushwork's pointing an agent at that instead of another consumer app.

Feels like the more interesting problem, honestly.

[Stefan Friedman @SteticIA]

Every AI roadmap this quarter has the same three bullets: coding agent, support agent, scheduling agent.

None of them have the bullet for a $220K CNC machine order that closes over a fax and a guy's word.

That's not a niche gap. That's most of what actually moves in the physical economy, sitting there untouched because it doesn't demo well in 30 seconds.

Strange that so much money is chasing so little of it.

[darkl0w @darkl0w]

@SteticIA banger

[darkl0w @darkl0w]

@chesny craaaazy

[darkl0w @darkl0w]

@SteticIA expedition gear vibe nailed by both, honestly

[Stefan Friedman @SteticIA]

KIMI K3 VS FABLE 5, SAME PROMPT OF AN EXPEDITION GEAR LANDING PAGE

same copy, almost the same layout

K3: $16.45. Fable 5: $37.94

nearly identical result

[darkl0w @darkl0w]

@LeoCreaIA fable 5 has that extra polish but k3 is closing in fast

[Leo @LeoCreaIA]

KIMI K3 VS FABLE 5, SAME PROMPT OF A SODA BRAND

same packaging concept, same visual tone

K3: $0.20 in API. Fable 5: ~$0.60

[darkl0w @darkl0w]

@chesny @LeoCreaIA the gap in type choices is bigger than expected

[Chesny @chesny]

EVEN ARCHITECTURE STUDIO LANDING PAGES ARE GETTING SOLVED WITH A PROMPT

→ K3: careful typography, clear visual hierarchy
→ Opus: more generic layout, less brand personality

K3: $0.20 in API. Opus 4.8: ~$0.40

design is getting solved by a prompt too

[darkl0w @darkl0w]

@chesny dinero fresco veo yo