José Valim

Creator of @elixirlang. Chief Adoption Officer at @dashbit, where we build and

[José Valim @josevalim]

It is finally here: Tidewave now supports Claude Code and OpenAI Codex.

Tidewave unlocks the full-stack potential of your favorite coding agent by tightly integrating it with your web app and web framework at every layer, from UI to database. More info 👇

[ElixirConf @ElixirConf]

We are thrilled to announce @smartlogic as a Platinum Sponsor for ElixirConf US 2026!

Creators of the Elixir Wizards podcast and experts in #Elixir development, they are true pillars of the community. Come meet the team in Chicago!

👉https://t.co/t8EsPImXvf

[ThinkingElixir @ThinkingElixir]

The final episode! After 6 years, we close with news on LiveView 1.2 colocated CSS, Earmark’s retirement, Sagents v0.8.0, #Postgres 19, @SpaceX buying @cursor_ai for $60B, and what’s next for David and Mark! @ElixirLang #ElixirLang https://t.co/gwciLQR6eR

[José Valim @josevalim]

Legit question: is this really beneficial?

I obviously understand the token-saving bits but that should not be the only thing to measure. Agents already frustrate me when they read only the minimum necessary of a file, leading to duplication, different coding styles, and so on. And we are often not measuring said impact when minimizing token consumption.

Not throwing shade on ast-grep outline at all. Doing it with tree-sitter and without an index is a great way to go about it, if that's what you need.

[Herrington Darkholme @hd_nvim]

ast-grep outline is released!

It gives you a fast local map of code structure, between grep and a full language server.

Syntax aware, no index to maintain, and extensible through declarative extraction rules.

[Code BEAM @CodeBEAMio]

Early Bird tickets for Code BEAM Europe 2026 are LIVE!

Secure your spot now to get the best price, here:
👉 https://t.co/mTHt3bfkFm

Don't miss Europe's top gathering for passionate BEAM, Gleam, Erlang, and Elixir developers.

See you in Haarlem!

#CodeBEAM #ElixirLang #Erlang #Gleam

[dex @dexhorthy]

love you all, this is hard work, there's still tons of unknowns, I'm trying to help you speed run to the frontier here. rant:

i'm sorry but if people are tokenmaxxing bugs into production and taxing your senior engineers with kLOC PRs that they didn't read themselves, those people shouldn't have jobs

coaching juniors on SWE fundamentals is hard work and it takes time but if you're not supporting the people trying to coach them, or actively undermining them by telling the team to "just build more loops" or "let go of reading the code", not gonna end well

[dex @dexhorthy]

it turns out if you do a lot of really good context engineering, you can get frontier intelligence for like half the price, and in some cases even better than a single frontier model

[José Valim @josevalim]

The duality of a man: getting excited about GLM 5.2 doing well as an open model AND deeply worried about this article from @Nature.

[nature @Nature]

Evidence suggests that AI-driven ‘deskilling’ is starting to happen in medicine, computer science and other fields

https://t.co/0xCHVAroP4

[José Valim @josevalim]

Ale was the main engineer behind Livebook Teams.

He implemented the client-server protocol (over protobuf), enterprise controls via OpenID Connect, and production monitoring + instrumentation. Strong recommendation for a backend engineer role.

[aleDsz @aledsz]

After 4 long years, unfortunately it's time to say goodbye to Dashbit! 🥲

Now I'm looking forward for new challenges, new opportunities and maybe new stacks?

Open to work with Elixir and other backend-focused stacks

[José Valim @josevalim]

@desireco I like this take a lot and that may be the conclusion we will arrive at. But even for that, we will have to measure!

[José Valim @josevalim]

Do any AI code review tools measure what humans catch that they miss?

Not just accepted/rejected/dismissed comment rates, but actual recall vs. human reviewers. If a human finds an issue the AI didn't, is that tracked, quantified, and fed back into future reviews?

I see a lot of comparisons between AI code reviews but we also to have measure them against the team's own baseline if we want to automate more parts of our processes. Some of those metrics also need to be quantified per directory/file.

[Hugo Baraúna @hugobarauna]

https://t.co/bftaI6Levm

[Ben Dickson @bendee983]

Very valid concern on the loop engineering trend. One concern I've had with all AI trends (vibe coding, tokenmaxxing, and now loopmaxxing) is the tendency to relegate thinking and design to a probabilistic system.

Loops have been around forever (e.g., every Windows application is basically an event loop), so it's only natural to wrap a loop around an LLM or AI agent.

However, what you put in that loop (and what you keep out of it) is very important. If you set a very vague goal and hope that the AI agent will figure out everything correctly, then you'll get unpredictable results (and waste a lot of money on LLM tokens).

If you design your loops carefully, using deterministic code for the deterministic parts, setting clear and verifiable goals, defining human oversight points, etc., you'll get a lot of value out of them.

Don't let loops become an excuse for laziness and sloppy design.

[José Valim @josevalim]

Since ChatGPT knows my name is José Valim (it is in the account info), it bias its replies towards Elixir, even when I want impartial discussions.

Memory is disabled too. I may need to rename the account to John Doe.

[José Valim @josevalim]

@hubertlepicki /goal is a banger. I am assuming that by, loop engineering, people are automating larger parts of the software development lifecycle.

[José Valim @josevalim]

Mixed feelings on the whole loop engineering thing.

On one hand, YES PLEASE. We have been automating the loop for decades, it is literally part of our jobs! More is good!

But I worry that we’ll go the wrong way and replace deterministic processes with uncertain ones.

[Charity Majors @mipsytipsy]

i wrote a post about the growing divide between AI skeptics and AI enthusiasts.

wins and costs are both real, but too often fall on different groups of people. when you're only experiencing half the story, it's too easy to write each other off.

https://t.co/qtjrOJA6a0

[José Valim @josevalim]

@juniorro16 O pior que não, eu estava na frente do sol! Esse deve ser o brilho natural meu, lol.

[José Valim @josevalim]

Anthropic has a narrow view of safety. Their actions in software engineering make me skeptical that they would act responsibly in positions of leadership in other domains.

[3by7 @3_by_7]

@josevalim Loops make the non-deterministic ones compound 🙈

[José Valim @josevalim]

@el_yawd I got it was a joke :D I didn’t notice it before and I thought it was still early enough to fix it! <3