Daniel Shapiro

Efficient frontier mentality // Applied AI Engineer @Blockworks // views my own

[Daniel Shapiro @_dshap]

The PA rn

(shorts safu)

[Daniel Shapiro @_dshap]

@ryeblocks You're saying this one case study is the source of the entire water source narrative? If true that is absolutely crazy lol

[Daniel Shapiro @_dshap]

Biggest takeaway from last few days is that it's pretty crazy that the majority of the population is anti-AI, anti-datacenter, etc.

As we all know success compounds super-linearly, and AI will only accelerate the gap between the people who are against AI vs those that adopt it and learn how to use it effectively

Not to mention its really not *that* expensive. For $200/m you get AGI. That's one date night nowadays

The internet, and now AI, are the largest equalizing forces in human history. And the people it could benefit the most are the most against it. Is what it is innit

[Daniel Shapiro @_dshap]

@MicahMuhr agreed - I thought this was a pragmatic take https://t.co/hWu1wEptXu

in terms of first point, we still don't know if it will replace humans or augment, so I think the point stands that you should try to take advantage of it now while we are still in the 'Centaur' phase

[Gavin Baker @GavinSBaker]

The governor-elect of Texas, Gina Hinojosa, has explicitly said that all currently permitted datacenters in Texas will be allowed to finish construction. This essentially covers all of the 2027 and likely well into 2028.

She is going to pass laws that ensure that all *newly* permitted datacenters going forward for delivery in 2028/2029 will pay taxes, commit to no increase in power prices for consumers, don’t impact water availability for people, be constructed with union labor and give “Texans a seat at the table.”

Sounds reasonable. She doesn’t want to lose a lot of good blue collar jobs and property taxes for small towns in rural Texas.

Datacenter commentary starts at 6:30 in this interview.

Also attaching her official campaign position from her website in the next post.

Sidenote: After watching a bunch of her speeches and interviews, I do think she is a political talent and probably has a future in national politics.

https://t.co/shCGNcdu9d

[Daniel Shapiro @_dshap]

@diesel_fan_S485 hard to disagree there. I like the new Texas governor's take on it, very pragmatic
https://t.co/FICasigFbw

[Gavin Baker @GavinSBaker]

The governor-elect of Texas, Gina Hinojosa, has explicitly said that all currently permitted datacenters in Texas will be allowed to finish construction. This essentially covers all of the 2027 and likely well into 2028.

She is going to pass laws that ensure that all *newly* permitted datacenters going forward for delivery in 2028/2029 will pay taxes, commit to no increase in power prices for consumers, don’t impact water availability for people, be constructed with union labor and give “Texans a seat at the table.”

Sounds reasonable. She doesn’t want to lose a lot of good blue collar jobs and property taxes for small towns in rural Texas.

Datacenter commentary starts at 6:30 in this interview.

Also attaching her official campaign position from her website in the next post.

Sidenote: After watching a bunch of her speeches and interviews, I do think she is a political talent and probably has a future in national politics.

https://t.co/shCGNcdu9d

[Daniel Shapiro @_dshap]

Very thoughtful response, thank you for this. And agree on many levels

I was more referring to the Democratic sweep and this type of polling data: https://t.co/0lSeb7BD9N

It seems that people are more anti-datacenter (due to resources) vs. anti-AI, which actually makes a lot of sense

[Violet @VioletFlame23]

Here's the polling data

50% of data center opponents listed resource concerns as one of their reasons for opposing data centers

Concerns about AI didn't even break 15%

(Oddly, 14% of respondents don't have *any* reason for opposing data centers? They're just going by vibes?)

[Daniel Shapiro @_dshap]

@Layton_Gott wya @thsottiaux

[Daniel Shapiro @_dshap]

BW on @OpenAI

[Blockworks @Blockworks]

NEW: Our official @OpenAI plugin is now live. Here's how to get started:

1. Head to https://t.co/wgYeArawxY and claim your free $100 of credits
2. Connect to the Blockworks MCP
3. Let your imagination run wild

Not a ChatGPT user? Our MCP works with any other agent as well.

[Daniel Shapiro @_dshap]

Best data in the industry now self serve!

Been using this MCP every day, rely on it heavily now for my trading

[Blockworks @Blockworks]

NEW: We’re opening up the Blockworks MCP to everyone.

Starting today, anyone can connect Blockworks to Claude, ChatGPT, Cursor or any MCP client. Sign in, claim your credits, and get your agents building within minutes.

https://t.co/wgYeArawxY

[Daniel Shapiro @_dshap]

One week out to @blockworksDAS London!

For anyone still deciding, can save some money with this code: DANIELS300

[Blockworks Data @BlockworksData]

One week until DAS London.
VIP: sold out. Allocator: sold out. GA: still open, but not for long.

Nov 10–11, Hilton Park Lane. Grab your ticket:
https://t.co/OkhmZG1rn0

[Daniel Shapiro @_dshap]

Congrats on the launch!

Now that claude is letting you use compute through other harnesses, super excited to try this out.

Have you heard anything regarding your claude usage limits going down faster if you aren't using your sub through the claude code harness? Curious if you have any deeper insight into this.

I absolutely love the codex/claude desktop apps (can't do CLI anymore) and the Factory app feels very close. If there was any easy way in the app to connect your codex and cc subs (just like in Amp Codes desktop app) that would be fire. I think that would get a lot more people to try out Factory.

Going to do some testing today on usage limits

[Daniel Shapiro @_dshap]

If you are wondering what happened today with the nance hack👇

[Yano 🟪 @JasonYanowitz]

1/ Everything you need to know about the Binance hack.

(written entirely by Claude using the Blockworks MCP)

A thread 🧵

[Daniel Shapiro @_dshap]

@shaundadevens hypershaunda

[Daniel Shapiro @_dshap]

Terra is not a “vibe coding” model

Unless you use the models extensively, it’s hard to get a good idea of what each class of model excels at

Although Luna 6 and Haiku 5.5 score well on these evals, they are simply not intelligent enough to be viable in production classification and reasoning tasks for general knowledge work

Terra and Sonnet are the least important/useful models in their respective lineups, but they still earn their keep IMO for a specific type of task

At the end of the day, I like having multiple options across the Pareto frontier. The more, the merrier!

(With that said I completely agree here I would never use terra/sonnet in Codex or Claude Code directly)

[Chubby♨️ @kimmonismus]

At least they are honest with the benchmarks. I get it, it’s cheap. But why would I use it when sonnet 5.5 is much better and even 6.1 Sol is almost the same pricing but better?

Feels like the weirdest release. Am I missing something?

[Daniel Shapiro @_dshap]

@fejau_inc Reflexivity at its finest

[Daniel Shapiro @_dshap]

Stoked for DAS London!

Specifically excited for a talk with OpenAI and Coinbase on AI, Crypto, and the State.

Use DANIELS300 for $300 off your ticket.

There’s also a £449/night hotel rate at Hilton Park Lane (usually £510), with limited rooms available.

Tickets + hotel details:
https://t.co/D4P9vp6VxW

[Daniel Shapiro @_dshap]

@radigancarter Thank you!

[Daniel Shapiro @_dshap]

idk what @thsottiaux just did, but after this reset my usage limits seem much more robust

Been running a few Astra threads and still at 98%

Is OpenAI back?

[Daniel Shapiro @_dshap]

@kimmonismus its impressive but unless performance is equivalent (relatively speaking) on other credible benchmarks, then its plausible that they overfitted/did benchmark-specific RL on terminal bench 4.0 vs ones like DeepSWE, Frontier Code, or CursorBench.

Check https://t.co/J1nGC6UPLN

[Daniel Shapiro @_dshap]

@thsottiaux where it at tho XD