Andrea Gurnari

💎 investing & building things at Moonfire 🇪🇺🇬🇧 Ex 2100, Atomico, Connect Ventures, GS, BlackRock. 🤖 🧠 Don't take yourself too seriously.

[Andrea Gurnari @agurnar]

I'm so keen to see positive destructive forces like @HugoAmsellem succeeding in finally pushing culture upstream of technology. let's go https://t.co/QWpaWujn7w

[Hugo Amsellem @HugoAmsellem]

https://t.co/X4DlWl8pEH

[Andrea Gurnari @agurnar]

@Konstantine @demishassabis really good interview @Konstantine - the story of Theme Park/SimCity as first AI sandbox (& The Sims & a 2023 Stanford study on the topic) is really wild. I wrote about it here for @Moonfire_VC just last week: https://t.co/kzPFwkmIx4

[Andrea Gurnari @agurnar]

In 1994, @demishassabis programmed 200-byte visitors for Theme Park. Need state. Decision rule. Environment design. The technical name for what he built is utility-based agent design. It is the pattern inside every agentic framework you are evaluating today.

[Andrea Gurnari @agurnar]

@demishassabis @TheSims @Stanford Full essay: https://t.co/BjkA0B1fsn

[Andrea Gurnari @agurnar]

The nostalgia we feel in AI is not childhood sentiment. It is pattern memory. You have been inside this architecture before. Most people have. Nobody told them what it was called. 🧵

[Andrea Gurnari @agurnar]

@demishassabis @TheSims @Stanford The architecture was never invented. It was worked out by anyone who tried seriously to make an artificial thing navigate a bounded world with human needs. It is still happening - in products nobody has thought to look in yet. The question is do you recognise it when you see it?

[Andrea Gurnari @agurnar]

such an obvious one thanks @maxleedev and congrats for your upcoming new role at OpenAI in 6 months

[maxleedev @maxleedev]

current chat interfaces suck, so i built a canvas for llms

[Andrea Gurnari @agurnar]

@tobdea peraltro verita assolute dette in 160 caratteri vs 3 ore di podcast. forse non siamo pronti

[Andrea Gurnari @agurnar]

Investing in AI in 2024 is an interesting game. You either find interesting technologies that are desperately looking for use cases, or use cases without any technological advantage.
Path to technology-use case fit becomes crucial to build winners.

[Andrea Gurnari @agurnar]

" (...) the ability to develop crisp mental models around the problems you want to solve and understanding the why before you start working on the how is an increasingly critical skill, especially in the age of AI."

https://t.co/1DYwnbxwIc

[Andrea Gurnari @agurnar]

Incumbents that leverage AI as a mean to a greater end (rather than making AI the end) create discrete (rather than continuous) innovations. The rest is going to be wiped out in a few years.

[Andrea Gurnari @agurnar]

@perplexity_ai is up to something. 👁️

[Andrea Gurnari @agurnar]

https://t.co/pZq3eSCdFx

[Andrea Gurnari @agurnar]

@stefanobernardi generational disadvantage

[Andrea Gurnari @agurnar]

@joeprkns @michuk Not in VC imho. Quality of data is not high+robust enough and long term success equation is too much based on error term than explainable. AI can be a massive help to gather + standardise + streamline process, but not to build conviction (ultimate alpha generator)

[Andrea Gurnari @agurnar]

@Thomas0tter @TextsHQ 🫡

[Andrea Gurnari @agurnar]

@pietrobezza @graywinpat The big question to me tho is: can you make a middle layer security feature (sms vs other auth methods) a premium feature? I feel users shouldn’t be asked to compromise on basic (or perceived as such) rights like security, esp if they don’t understand the cost behind a feature.