All the Elixir libraries I’ve been publishing recently and OpenPencil are part of the same vision. The idea is much bigger than just building another design harness.

Finally deploying Exograph

The music recommendation systems built around that principle are even more annoying. Fine. I’ll do it myself.

I feel like I’m living in my childhood dream. I talk to a terminal in plain English, delegate all the boring work, and have enough power to fix every imperfection in the virtual world that used to annoy me.

If you’re interested in the technical side of what I’m building and how all the pieces fit together, I’ve dumped most of my thoughts here: github.com/elixir-vibe/build… It doesn’t really cover the product side of the vision. I’ll hopefully publish that later this week.

What if Elixir Vibe coding agent and pi had a baby? Coming soon.

Переварят ли доставки столько новых курьеров?

I’ve talked about this business model privately many times. Okay, you don’t want to burn VC money subsidizing tokens anymore. Fair enough. So you ask users to pay separately for the LLM and for the harness. But then an open-source alternative emerges. Maybe it’s OpenPencil. Maybe it’s Open Design. Maybe it’s something else entirely. It may not be as polished, but it’s free and has a community around it. What is your next step? Any extra feature you ship will be copied in no time.

I was able to get roughly 2× the indexing performance in Exograph, my Elixir code search engine, by switching the default backend to DuckDB. Search got a lot faster on several important paths too. Yes, I know PostgreSQL can be extended with things like ParadeDB for full-text search, Citus for distributed Postgres and columnar storage, and all kinds of other serious machinery. But in practice, those setups are more operational complexity than I want here. My DuckDB-based setup is much simpler to operate, and I have full control over it through sweet Elixir APIs. So I’ve finally solved one of the two major blockers for the OpenPencil backend (yes, you heard that right — there will be a backend soon) and its companion platform. I think I’ll be ready to tease a few more building blocks this week, followed by a public roadmap for the whole ecosystem. Stay tuned.

Or fine-tune your model to do more stuff autonomously (if you work at OpenAI)😅 I’ve been building all my agentic workflows around this simple idea from the very beginning.

Yeah, this is how much I’m afraid of @badlogicgames and his contribution guidelines now: github.com/dannote/dot-pi/bl…

People of Pi! I’m currently in the self-reflection phase of my vibe-coder life cycle, so I decided it’s time to turn all my current coding tricks and habits into tools again. Unlike some vibe coders on Twitter with Mac Mini stacks and 10-screen setups, I like controlling everything that happens and owning the generated code. I’ve been using this setup for a while, and maybe some of that experience will be useful to you. The setup has always been open: github.com/dannote/dot-pi (Most of the mess I produce with it is open as well) I use a collection of custom tools for web and documentation lookups, worktree management, and many other things. They’re not perfect, but that’s the reality I actually live in. Today I spent a lot of time polishing them. Soon there will even be a tutor extension that will help you turn into me. (I strongly discourage doing that.)

I can spend 10x usage limits easily 😅

I I tried a lot of different packages, but eventually gave up and implemented it myself 😅 More details in the README: github.com/dannote/json_code…

Working on something similar, but much more flexible and built for Elixir/Phoenix. Jido-based. Both embeddable and runnable as a standalone proxy (similar to LiteLLM). ETA: next week.

QuackDB 0.5 is out 🦆 I’m increasingly convinced that @duckdb is becoming one of the best storage choices for hobby to medium-sized projects. Postgres is still the obvious default for highly concurrent OLTP systems, but many small and medium products are read-heavy, append-heavy, analytical, or operated by a small team. For those workloads, DuckDB’s columnar execution model, vectorized engine, rich SQL surface, and embedded/local-first deployment model can be a much better fit. I intend to use DuckDB as the primary storage layer for the platform I’m building, so QuackDB focuses on the parts I need in a real Elixir app: supervision, connection pooling, Ecto, fast append paths, dataframes, telemetry, and helpers for DuckDB’s analytical SQL. 0.5 focuses on making the Ecto and analytical side much more complete: - append through Ecto with defaults and returning - using an Ecto repo directly with QuackDB native query/append APIs - advanced join patterns, including semi/anti and ASOF-style queries - DuckDB star/COLUMNS expressions in SQL and Ecto - broader LIST/MAP/STRUCT helpers - PIVOT, UNPIVOT, GROUPING SETS, ROLLUP, and CUBE builders - LIST lambdas in Ecto with `fn` syntax and `case_when` - better nullable/schema type handling for append-heavy workloads QuackDB also continues to cover the broader integration surface: supervised DuckDB, DBConnection/Ecto, native append, Explorer dataframes, Table.Reader results, Geo/WKB, telemetry, managed DuckDB binaries, and DuckDB-specific SQL helpers. github.com/elixir-vibe/quack…