Aayush Giri

snr devrel @Nethermind @NethermindStark // ETH maxi // solidity + rust onchain infra & perps // prev @spheronFDN // code hard, shitpost harder

[Aayush Giri @AayushStack]

devrel travel looks glamorous from the outside, you see pics like this and think “damn, nice life”

but here’s the thing, this job requires you to talk. a lot. workshops, talks, twitter spaces, 1:1s with devs, answering the same questions with the same energy every single time and it only works if you genuinely love it. like actually love explaining how things work to people because if you don’t, you’re just performing enthusiasm for a paycheck. and developers can smell that from a mile away the perks are real. the views are beautiful. but so is the energy it takes to stay “on” when you’re exhausted as

uncle ben once told my friend spiderman - with great power comes great responsibility
this has to be a passion. otherwise you’re just working to work.

[Aayush Giri @AayushStack]

i don't have a five-year plan, i have a five-tab problem.

[Aayush Giri @AayushStack]

the bug was in production, in plain sight, wearing a passing test.

[Aayush Giri @AayushStack]

i love rust the way you love a friend who's always technically correct.

[Aayush Giri @AayushStack]

the AI-and-crypto convergence everyone keeps drawing on whiteboards is real, but it's not the convergence in the pitch decks. it's not "AI tokens." the actual convergence is a problem meeting its solution across a fifteen-year gap.

crypto spent fifteen years, mostly mocked, building one specific thing extremely well: machinery for letting an untrusted autonomous process act in the world with verifiable, accountable boundaries. signatures, proofs, deterministic execution, "show me, don't tell me."

for fifteen years the honest critique was that this was an answer without a big enough question - the autonomous processes were just simple scripts. and then AI handed crypto the question. agents are autonomous processes about to touch real money, real identity, real consequences - and "trust the lab" does not survive contact with that.

you cannot have an agent move funds on a vibe. you need it to act inside boundaries it cannot exceed and prove what it did after. that's not a new thing someone needs to invent. crypto already built it.

it was just early - early to the point of looking pointless. the convergence isn't two industries merging. it's one industry finally being the answer to a question the other one just finished asking.

[Aayush Giri @AayushStack]

the weirdest skill the last year taught me is reading an agent's work the way an editor reads a draft - fast, suspicious, scanning for the confident sentence that's quietly false.

it's a real skill and it's not the same as writing code. it's closer to taste, to smell, to that itch when something is too clean.

i think "can you tell when the machine is wrong" is about to be the most valuable line on a resume and the hardest one to fake.

[Aayush Giri @AayushStack]

the contract is immutable; my regret about it is not.

[Aayush Giri @AayushStack]

open source is the only place where giving it away is the growth strategy.

[Aayush Giri @AayushStack]

i ask the agent for one function and get a small opinionated framework.

[Aayush Giri @AayushStack]

gm! gm!

[Aayush Giri @AayushStack]

"we're hiring" is sometimes a status update, not a job.

just saying.

[Aayush Giri @AayushStack]

the thing about watching a project hit #1 on github trending is you realize how thin the line is between "obscure repo" and "everyone's talking about it." it's not gradual. it's a phase change.

one week nobody, next week everybody. and it teaches you something about your own unlaunched ideas: the world isn't slowly evaluating you and finding you wanting.

it just hasn't noticed yet. those are very different problems, and only one of them is fatal.

[Aayush Giri @AayushStack]

AI-native is the most abused phrase in tech right now so here's the test i actually use:

remove the model. if you still have a coherent product, you built a normal app and taped a copilot on. if the whole thing collapses into nonsense, congrats, it's AI-native.

openhuman isn't AI-native because it talks to you in a cute mascot. it's AI-native because "ingest two years of your gmail, compress it 80%, and rebuild a memory tree every 20 minutes" is a sentence that was physically unbuildable three years ago.

the model isn't a feature bolted to the side. the model is the only reason the architecture exists. that's the line. most things claiming to be on the far side of it are not.

https://t.co/qVaArrFTPY

@tinyhumansai @senamakel

[Aayush Giri @AayushStack]

"MVP" stands for "my patience, eventually."

[Aayush Giri @AayushStack]

late thought, posting it before i lose the nerve. i've spent six years building in two industries - crypto, now AI -that are both, in their honest moments, attempts to answer the same question: how much can you trust a process you didn't personally run?

crypto answered it for money and computation. AI is being forced to answer it for cognition and judgment. and i've started to think that question isn't really a technical one at all.

it's the oldest human question wearing a hoodie. how do you trust what you can't fully see. we've been answering it with institutions, with reputations, with law, with faith, for all of history - and now we get to answer it with math and code, which is new, and which is the part that keeps me here.

i didn't get into this for the tokens or the model benchmarks. i got into it because i get to spend my actual working hours on a problem, using tools that didn't exist when i was in college, from a desk in India, shipping into the same internet as everyone else.

it doesn't get easier and it doesn't get certain. it gets more interesting. that turned out to be the trade i actually wanted.

[Aayush Giri @AayushStack]

gas fees are just the network charging admission to your impatience.

[Aayush Giri @AayushStack]

i think the most honest description of where software is heading is this: we are moving from building products to growing systems. a product is a thing you finish, polish, freeze, and defend. a system is a thing you start, feed, supervise, and let change shape under you.

openhuman isn't really a product in the old sense - it's a memory that grows, a loop that runs, an agent that's slightly different next week because your week was different. that's a system.

and i think most interesting software is going to feel like that soon: less like an object you bought and more like a garden you tend, with the agent as the thing that keeps growing while you sleep.

it changes what "shipping" even means. you don't ship a system once. you ship the conditions for it to keep becoming. i find that genuinely beautiful and slightly unsettling, which is usually the exact mix that means something real is happening.

[Aayush Giri @AayushStack]

the testnet is where confidence goes to get fact-checked.

[Aayush Giri @AayushStack]

every "final version" is a draft that got tired.

[Aayush Giri @AayushStack]

here's what my years in this industry actually taught me, stripped of the motivational coating: nobody is coming, the map is wrong, the experts are guessing, the timeline lies, and the only reliable signal is what you can get a real person to actually use.

that sounds bleak. it isn't. it's the most freeing thing i know. if nobody has the answer, you're allowed to just go find out. that permission was always there.

[Aayush Giri @AayushStack]

gm!

[Aayush Giri @AayushStack]

i build at night because that's when the internet stops giving me opinions:)