d0100

I've reduced queries from minutes long to 2s in Postgres by duplicating a CTE and keeping it unused

Query planner feels pretty LLM-esque already

LLMs seem to create abstract, local jargon as a side effect of way it reasons using tokens

ChatGPT told me its "semantic compression"

Gamers Nexus has a 2h video doing reverse engineering and a lot of tests

We can always streamline creativity by allowing agents to post jobs on mechanical Turk

> a self-contained, signed extension ecosystem

What does this mean? Can I easily sign my own private extensions with my own key?

We are currently allowing unsigned extensions, loading them, then locking the duckdb config to prevent loading unsigned/untrusted extensions

> we hire Juniors to help fill the talent pool with the hope that they will be productive in a few months

The sad reality for smaller companies is that we cannot compete salary-wise, which means training juniors that will leave the minute they become productive

But mostly the issue is with the market, there are only so many things you need a junior to do in a body shop that is getting squeezed

Other companies are also cash constrained and need to pivot fast to "AI"-something, which juniors wont really help with on the short-term

An employee just got phished by adding a number to a legitimate deviceAdd login route that bypasses 2FA and adds a device with full access to office and mail

Probably working as intended...

My workflow for production code:

1. Pick the most complete project boilerplate (fullstack JS can easily introduce security bugs, SPA + API is best as cheap linting solves most problems)

2. Project skills (how to CRUD without mess)

3. Use worktrees for concurrent features, local session for conflicts

4. Local session for QA and refinement

I use Copilot and GPT 5.4

Managed to shorten pre-AI priced ongoing projects to 2 weeks or a month

> GPT 5.4 instead of 5.5.

I am exclusively using 5.4 because its only 1x and very good, but the github calculation showed my once $40 become a $680 billing

That is too expensive and not worth paying

> Python is, thus, clearly superior to other languages in terms of ease of review.

Do we get visual comparisons along with this bold claim?

I remember doing a MailChimp "clone" in Laravel some 12 years ago and implementing an email builder using their templates

I am using TS sandboxed in deno for all our agent code generated from a UI builder (inspired by OpenAI's own agent builder, and spits out the same code output)

Are you using Temporal with distributed workers?

We have a simple worker setup and temporal is pretty easy to setup

Out only issue is really needing an intermediary data store for task result storage

We are using DBOS in new projects as it's even simpler and the downside (task log interface behind saas) is easily remedied with a copilot generated task viewer

Tried using Zed but for some reason the AI can't open the browser?

The trick is to constrain the LLM to program in a very defined coding style

If I make the LLM generate code that follows my own code architecture and style, that should be enough creative input