Chris McCord @chris_mccord

Creator of the Phoenix Framework, Author of Metaprogramming Elixir, Programmer at @flydotio, builder of things chrismccord.com Charlotte, NC Joined February 2011

Tweets

6K
Followers

23K
Following

394
Likes

3K

ThePrimeagen @ThePrimeagen

2 days ago

after one month of using elixir i can confidently say that i really like the language

49 37 1K 75K 27

Introducing Tidewave Web for Rails and Phoenix: a coding agent that runs in the browser alongside your web application, with full page and code context. Tidewave deeply integrates with your stack, from the database to the UI, making AI development more seamless and efficient.

23 121 667 80K 268

Download Video

José Valim @josevalim

2 weeks ago

We asked Tidewave to improve its own homepage with autoplay of videos as we scroll through our features list. It implemented the feature and used contextual browser testing to scroll through the page, check the DOM, and verify it works! 🎯 Coming for Rails and Phoenix... today!

5 30 269 19K 60

Download Video

Chris McCord @chris_mccord

3 weeks ago

LLMs are surprisingly good at predicting the number of lines they are about to write, across wide range of line density, ie markup vs code. Seems like an emergent property and all SOTA models can do this within spitting distance of true thing. Great for progress bars

2 1 15 3K 1

Chris McCord @chris_mccord

4 weeks ago

gpt-5-mini is oddly on par with gpt-5 for a lot of agent use, sometimes seemingly better at basic instruction following. Great pricing but unfortunately both require high reasoning for good output which is super slow, & both miss the mark for me compared to comp, & even gpt-4.1

4 1 35 4K 3

Chris McCord @chris_mccord

4 weeks ago

I guess I'm still stuck in not-smart mode:

5 0 25 3K 0

Download Image

Chris McCord @chris_mccord

4 weeks ago

It's possible gpt-5 reasoning_effort is producing slightly better results, but at like 5x the cost in speed. Meanwhile sonnet outperforms by massive margin in speed, instruction following, and quality with thinking disabled

1 1 18 2K 1

Chris McCord @chris_mccord

4 weeks ago

Here's @grok 4 with the same failed gpt-5 genserver + DB based chat prompt. It got to a working app and realized it needed to restart after sup tree change. Design could be better, but actually works.

2 1 13 2K 0

Download Image

Chris McCord @chris_mccord

4 weeks ago

I triple checked that I didn't typo the model. I've been too spoiled by all the amazing releases over the last year, so maybe miscalibrated expectations on my side. As general model/multimodal it's likely major step up, but for code seems to underperform gpt-4.1 (for elixir)

2 1 17 2K 1

Chris McCord @chris_mccord

4 weeks ago

Claude with same exact "fix all the Todos.list_todos by making real queries to efficiently fetch and count things" prompt put things into a single query and wrote something reasonable: x.com/chris_mccord/s…