SHELDON EVANS

investor. techno optimist. life extension maximalist. making tech work harder so i don't have to.

[SHELDON EVANS @SheldonEvans]

How to build an AI company in 2026:

> raise money
> build the exact same agent everyone else is building
> vacuum user data
> sell it
> Series B

[SHELDON EVANS @SheldonEvans]

We’re entering the real super ( $HYPE ) cycle.

Probably gonna power trading on X is my assumption.

[SHELDON EVANS @SheldonEvans]

@martypartymusic Idiot it’s called sarcasm

[SHELDON EVANS @SheldonEvans]

@make_robot_go @roboticscorp Robots are great but this looks depressed and creepy.

[SHELDON EVANS @SheldonEvans]

@LawtonAver58308 @Tesla Okay stay sad and boring

[SHELDON EVANS @SheldonEvans]

The @Tesla Roadster demo has illustrated yet again that a lot of people online seem to have forgotten how to be impressed by something that is, objectively, fucking cool.

Telling someone in the 1800s that micro-explosions would replace the horse and that we’d teach rocks to think would’ve sounded insane too but that’s how we got cars and computers.

If your horizon is only two days long, then yes, this is just a cool party trick.

It’s a flying car. It’s fine to say it’s cool af and will probably lead to more cool shit for humans and society.

[Skunk Chubberly @SkunkChubberly]

@SheldonEvans @abcdedcba69 @Tesla You think that this tech comes with mass-consumer use, air-parking and a complete remodel of the highway system… and you’re wondering why it’s a bad take?

Air thrusters on a car is cool. That’s all it is. And all it has the capacity to be for the foreseeable future.

[SHELDON EVANS @SheldonEvans]

@abcdedcba69 @Tesla I’m not saying the current version of this but it’ll obviously improve from being just a party trick and get to the point where it’s absolutely usable. How is that a bad take?

[SHELDON EVANS @SheldonEvans]

I believe most people underestimate technology like this because they have non-creative minds. @Tesla doesn’t seem like real life at first because…

First-order thinking: "We don't need rockets on a car."

Second-order thinking: boost out of a crash before it happens. Multi-level highways without building onramps, boost to lvl 2/3/4. Park outside your window in an apartment block. Literally even just rocket jet some loser trying kidnap your family members. So many use cases it's insane.

We live in the greatest timeline to experience tech like this.

What's the first thing you'd use it for?

[Tesla @Tesla]

Real life, baby

[SHELDON EVANS @SheldonEvans]

@78Xantus @Tesla You’re assuming technology doesn’t get better over time. What is this dumb take? Of course it will eventually round to 100% accurate and safe.

[SHELDON EVANS @SheldonEvans]

Happy 10/10 to those who celebrate.

Liquidate all bears this time please.

[SHELDON EVANS @SheldonEvans]

While everyone is freaking and getting into “bunker mode” with their crypto over the possibility that ECDSA gets cracked in the next few months with ai, we must remember that legacy systems and fiat aren’t safe either.

Some dude just vibe hacked South Korea’s biggest banks with not even the best models.

I believe we’ll see all of our systems break in the coming months and years and we’ll need radical architecture and system redesign.

We’ve entered an era of cat and mouse and it’s only going to get faster, crazier and more interesting.

What breaks first is anyone’s guess unfortunately.

[Andrew Curran @AndrewCurran_]

Last week some of South Korea's biggest banks were hit by a cyberattack. Thanks to a report from CrowdStrike tonight, we now know the entire hack may have been done by a single person. He used a combined stack of an open-source AI penetration tool named ARTEX, DeepSeek v4.1-Flash, GLM-5.3, Grok 4.6, and Claude Code.

[SHELDON EVANS @SheldonEvans]

@Robhay3s who gaf, im a tech person and it makes my day better

[SHELDON EVANS @SheldonEvans]

Elon wins the ai race with this imo.

No one really cares what the underlying model is as long as you get a successful outcome from the tasks you’ve assigned to your agents.

Whatever it takes under the hood is irrelevant , which is already the case for most tech.

I, like many others, jump ship constantly to whichever model works best at any given time and often MOE wins.

@bot has the distribution and that’s the only thing that matters as we approach ASI and the average success rate of any task assigned to any frontier model approaches 100%.

[Elon Musk @elonmusk]

Important note regarding Grok @Bot:

Going forward, @SpaceX will use the best back end model for any given task, including Claude Opus 5.5, MidJourney, Suno and other leading APIs.

Whatever is most likely to give you the best outcome.

[SHELDON EVANS @SheldonEvans]

@thomasjeans Why? Explain please

[SHELDON EVANS @SheldonEvans]

@ZeroMazed They’re not going to run out of runway

[SHELDON EVANS @SheldonEvans]

@m4rtin_georgiev Stuck in a terminal

[SHELDON EVANS @SheldonEvans]

@brantcraft It will be exactly like this. No one will care nor will it really matter in a post AGI world I think

[SHELDON EVANS @SheldonEvans]

I mean with every model it gets cheaper and cheaper to complete the tasks simply because of fewer failures. So it’s probably not a bad idea for them to eat the token cost as we’re edging to ASI. Sure maybe a year ago we couldn’t extrapolate accurately how fast we’d get to models of this capability but now it’s pretty clear we’ll reach ASI sooner rather than later.

[SHELDON EVANS @SheldonEvans]

@claudeai Make it 100% and we have a deal

[SHELDON EVANS @SheldonEvans]

@levelsio The problem is the algorithm rewards controversy/ragebaiting/negativity so they’re increasingly incentivized to just have the opposite take on everything. Seems to happen on all platforms lately.