Programming in Elixir without AI - Web Crawler

For anyone who still wants to learn to code…

https://www.youtube.com/watch?v=2fOg_ieOEEk

Hologram v0.9: Realtime and More

Realtime comes to Hologram - v0.9 is out! Your server can now push updates to connected clients with no polling: broadcast an action, and Hologram runs the matching handler on every subscribed client, all in pure Elixir. The most complex feature shipped so far - plus the with special form, AI assistant support, and a new mix holo task. More here: https://hologram.page/blog/hologram-v0-9

Implementing a GPT model from scratch

This post implements a GPT-style model from scratch in Elixir/Nx, building on the attention mechanism post. We start with the GPT-2 configuration and parameter counts, then add layer normalization, GELU feed-forward layers, shortcut connections, transformer blocks, the full GPTModel, and a simple greedy text generation loop. https://karlosmid.com/2026/05/implementing-a-GPT-model-from-scratch/

Thinking Elixir 305: Eleven Minutes to Mayhem

Episode 305 of Thinking Elixir. News includes Elixir 1.20.0-rc.6 arriving as likely the final release candidate before v1.20.0 ships, completing a ~15-week roadmap and delivering full type inference across applications and dependencies. The EEF 2026 election results are in with 3 returning and 1 new board member, LiveStash v0.3.0 lands with a Redis adapter and auto-stashing for Phoenix LiveView state recovery on WebSocket reconnects, a call goes out for BEAM ecosystem companies to step up and support the EEF’s critical security work, and GitHub suffered a significant breach when a hijacked VS Code extension quietly exfiltrated ~3,800 internal repositories in just 11 minutes, and more!

https://www.youtube.com/watch?v=UWFn3oavR3s

Is the Erlang Ecosystem community (Elixir, Erlang, Gleam) paying attention to what’s happening in the world of ‘AI’?

I’m not sure if I’m not paying attention, or if it’s just not getting through to me, but I don’t see much discussion within the Elixir, Erlang and Gleam communities about the issues surrounding the use of ‘Artificial Intelligence’ (AI) and, in particular, Large Language Models (LLMs) in software development.

I see major problems—primarily ethical and environmental, but also financial (in the long term)—with using LLMs in software engineering.

In this post a provide a few links for those that are interested in learning more:

https://dev.to/adolfont/is-the-erlang-ecosystem-community-elixir-erlang-gleam-paying-attention-to-whats-happening-in-2mkb

Long — a self-hosted LLM agent on Elixir

https://github.com/mjason/long

Long is a self-hosted LLM agent runtime built on Elixir/OTP. Point it at any provider (OpenAI, Anthropic, Google, DeepSeek, local Ollama — 20+ via ReqLLM) and talk to it through a built-in web UI: it runs a ReAct loop with tools, keeps four tiers of memory, schedules recurring tasks, searches and reads the real web, and answers your WeChat and Telegram DMs — all from a single process.

What sets it apart is that the agent’s entire capability layer is one introspectable GraphQL tool over its own data, instead of a hand-maintained pile of bespoke tools — the model already speaks GraphQL, so it discovers and operates its own memory, schedules, and sessions on its own.

And it installs like a CLI tool, not server infrastructure: one self-contained binary (the Erlang VM is bundled), everything under ~/.long/, no external services to provision (just SQLite + the filesystem), userspace autostart, LAN-first defaults — so getting it onto a spare Mac mini or Linux box is curl | bash plus an API key, and it runs like an appliance from there.

GenServer State Management in Elixir: A Production Order Book

What lives in a GenServer’s state, what gets persisted, when to reconcile, and why the broker — not the database — is the source of truth.
https://seriousalchemy.com/genserver-living-order-book

Submit a talk for Code BEAM Europe 2026

Hello everyone,

This is the reminder, that you can submit your talk for Code BEAM Europe 2026, 21-22 October in Haarlem, Netherlands.

We are looking for 20- or 40-minute in-person talks. This is an opportunity to showcase your BEAM (Erlang, Elixir, Gleam) solutions, products, or technical insights.

For 1st time speakers: We highly encourage you to apply. We offer full mentorship, including direct help with evaluating your idea and preparing your slides.

Topics we are looking for:

• Data Sovereignty & Digital Autonomy
• Open Source
• Modern Architecture (Backend & IoT)
• Growth & Adoption
• Real-World Production Realities
• Scalability & Sustainability

Deadline: 31 May Submit here: codebeamstaockholm.com#cft Contact: info@codesync.com

We look forward to your proposals!

Thinking Elixir 304: Types, CVEs, and Hot Reloads

Episode 304 of Thinking Elixir. News includes a major milestone for Elixir’s set-theoretic types as inference of all language constructs is completed and merged with Elixir v1.20.0-rc.5 hot on its heels, OTP 29.0 drops as a major release with secure-by-default SSH, post-quantum SSL key exchange, Erlang doctests, and more, a wave of high-severity CVEs hits the Elixir and Phoenix stack prompting the EEF CNA to take on a larger work load as AI-driven vulnerability reports surge, string processing in Elixir gets a serious speed boost via SWAR (SIMD Within A Register) optimizations with 1.5–5x improvements across Base and String operations, and a handy tip for enabling state-preserving hot reloads in Phoenix LiveView with just a small dev.exs config tweak, and more!

https://www.youtube.com/watch?v=32QmElA8H0I

ElixirGtBridge: A next generation elixir IDE

Want a new Elixir IDE that lets you make your codebase more explainable? Want to be able to more easily explore and edit your elixir applications? Want to make custom guis for all parts of your system while it’s live? Then look no further.

See the v0.17.0 version of the Elixir GtBridge below

https://www.youtube.com/watch?v=awulq0-VOtA

Hex | GitHub | GlamorousToolkit

Mold - a tiny, zero-dependency parsing library for external payloads

Mold parses JSON APIs, webhooks, HTTP params and other external input into clean Elixir terms - coerces types, renames keys, checks structure, and returns {:ok, result} or {:error, errors} with traces.

Cheatsheet | Documentation | Hex | GitHub

You're probably underusing middleware for HTTP response handling

A note on a pattern I keep running into in Elixir API clients: status-code checks and body parsing live in code after Tesla.request, instead of inside the middleware chain itself. It looks fine from the caller’s side, but it silently breaks observability. A 500 translated to {:error, _} looks like a successful request to Tesla, so :telemetry and anything else wired up to the request lifecycle stay quiet.

https://dev.to/arturplysiuk/youre-probably-underusing-middleware-for-http-response-handling-1djf

How do I write Elixir tests?

Personal guides for writing tests that are readable and maintainable. Stuff to use, stuff to avoid, and how to organize stuff.

Furthermore, I think that mocking must be destroyed.

https://hauleth.dev/post/writing-tests/

Concurrency, understanding the BEAM limits - Lorena Mireles Rivero [webinar]

https://www.erlang-solutions.com/webinars/concurrency-understanding-the-beam-limits/

The BEAM makes concurrency powerful, but resilient systems come from understanding capacity limits and designing around them.

Watch Lorena’s webinar where she breaks down how schedulers, mailboxes, and latency behave under load, and explains approaches like workload distribution, back pressure, and graceful degradation. All that with real-world examples.

FastDecimal 1.0 — a pure-Elixir alternative to decimal, ~11× faster

FastDecimal is a pure-Elixir decimal library aimed at workloads where the arithmetic itself is the bottleneck (ledgers, pricing, batch sums). Geomean ~11× faster than decimal across add/sub/mult/div/compare/parse on BEAMAsm — see the benchmark table for the full run.

Migration is one line for typical code: alias FastDecimal.Compat, as: Decimal. Full feature parity with decimal except Decimal.Context (intentional — precision is per-call).

Hex: https://hex.pm/packages/fastdecimal · Docs: https://hexdocs.pm/fastdecimal · GitHub · Migration guide

Updates to Permit: Igniter setup and better defaults

The Permit authorization suite has just got a batch of updates with Igniter-powered generators for Phoenix, Ecto and Absinthe. Read Dawid’s release notes article at the Curiosum blog.

Thinking Elixir 303: The Taming of the Slop

Episode 303 of Thinking Elixir. News includes the Erlang Ecosystem Foundation publishing its 2026 board election candidates with voting now open, a new GitHub organization called Elixir-Vibe launching with tools to detect and fix AI-generated Elixir “slop” — including ExSlop and the related semantic linter Credence, erlang_python 3.0.0 arriving with true parallelism by embedding CPython into the BEAM as a first-class citizen, ElixirConf EU 2026 videos beginning to drop including keynotes from José Valim and Chris McCord, and more!

https://www.youtube.com/watch?v=t-VuB97ZY2c

Reflecting on Event Sourcing ElixirConf

I talked about Event Sourcing in August and it just went up on youtube. It was a nice chance to reflect on what’s changed in the last 6 months and what I’m appreciating now. https://sylverstudios.dev/blog/2026/05/12/event-sourcing-elixir-conf

Lockstep — find race conditions in BEAM code and replay them on demand

Lockstep is a concurrency test runner for the BEAM. You write what looks like an ExUnit test; it runs many times under controlled scheduling, and when it finds a race it gives you the exact schedule to replay.

Real bugs surfaced building it: Phoenix.Tracker pre-fc5686f, NimblePool pre-e18f45f, Hammer’s Atomic.FixWindow.inc/4.

Features:

• 5 scheduling strategies + trace replay + counterexample shrinking
• Drop-in OTP wrappers (GenServer, Task, Registry, Supervisor, ETS, Atomics, …)
• AST rewriters for Elixir + Erlang — works on unmodified upstream source
• Multi-node Lockstep.Cluster with partition / heal
• Optional Anthropic Claude integration to explain failing traces

https://github.com/b-erdem/lockstep

TelegramEx – make Telegram bots with ease

Built telegram_ex to make Telegram bot development in Elixir less painful.

Macro-based API handles boilerplate for you.

Features:

• Text, photos, documents, stickers, video
• Inline & reply keyboards
• Stateful FSM with routers
• And more…

Would be appreciate your feedback!

https://github.com/lsdrfrx/telegram_ex