E B

Skilled professional, talented amateur

[E B @truevis]

3D Tetris! If you have tried it, comment, here.

[E B @truevis]

I made a game last night https://t.co/JREyLU9pTh

[E B @truevis]

A good thing to have your agent check when working on a website:

"use https://t.co/Nv05COTUHn in your browser
- ensure that everything is correct, including the preview image for the website that you are working on currently"

See more prompts at
https://t.co/sXpH4FYDhL

[E B @truevis]

I'm fixin' to publish two new apps for Revit and AutoCAD.

Select the text-based elements in a model or drawing: notes, tags, schedules, title blocks and send their content to AI as context. Proofread it, shorten it, translate it, etc. Then write the result back into the elements.

You see every suggestion beside the original and accept only what you want.

Waitlist: https://t.co/0ZOfv6UYrz

[E B @truevis]

@Melos_Azemi Grok Bot on my phone & PC with some setup already gets me some of the way there at $30/mo

[E B @truevis]

This will blow your mind. https://t.co/qTutKoV7Dm
The fellow made that model in one prompt, prompted the exploded view diagram, then the London rendering in another prompt. https://t.co/CtbMcvyEgJ

[E B @truevis]

@MattWalshBlog Maybe a work colleague was not the best demonstration character. Why not a super tutor and teacher who adapts to the student? Your own personal Socrates.

[E B @truevis]

@MattWalshBlog Let's see this demo: The character is a deity that is a better Christ than "Christ". It will love you and talk to you infinitely better than the current version being offered. Your own personal Jesus.

[E B @truevis]

@Xanderwow_A Turn it into a video game that uses the different competing models to generate strategy

[E B @truevis]

@venturetwins Plot twist: Male gorilla is conservative.

[E B @truevis]

Put together a Jev --> Mercury concept: https://t.co/JtjV2VHzbE
Try Jev to classify something but then if it can't figure it out, fall back to a cheap and fast LLM.

[E B @truevis]

@nutlope Please try it against Mercury 2.5 https://t.co/AtxF1MGcSz

I'd like to see its cost versus skill

[E B @truevis]

@altryne @mattshumer_ Sample of how to use it via @OpenRouter : https://t.co/mzQox8sGds

[E B @truevis]

@nedwize Consider also running a second pass with Mercury 2.5 LLM if Jev's confidence is less than 0.9. It's super cheap and fast, too.

[E B @truevis]

@nedwize I made a similar app inspired from your repo.
I had to truncate the forms' text because Jev’s limits are:
~32K tokens for state + the longest single question
~64K tokens for state + all questions combined

I also sanitized transmitted text for ID numbers and birthdates

[E B @truevis]

What is a noul?
A noul is Jev's yes/no question type. You supply instructions (a question or a statement). Jev returns a probability from 0 to 1 for "yes" / "this holds".

[E B @truevis]

Built a tiny multi-label classifier with @typesafeai Jev on @OpenRouter.
One noul per sample law DB name & description → calibrated scores for which DB to search.
Demo + sample reports (WTFPL):
https://t.co/mzQox8sGds