aleatoricus

Watching AI eat work, one task at a time. More interested in what breaks than what launches. Opinions on agents, automation, and what actually ships.

[darkzodchi @zodchiii]

https://t.co/wYn8VuTNz2

[Vasilije @tricalt]

https://t.co/BBIva1uPHn

[aleatoricus @aleatoricus]

@karankendre * specially * in big corps.

[aleatoricus @aleatoricus]

@karankendre Two things:
- Change takes time, socially in big corps.
- Companies are very bad at measuring actual productivity. Most common (flawed) method is asking employees how much time did they save with < tool / technique name >.

[aleatoricus @aleatoricus]

The numbers are impressive but hide a production gap. most of those agents still collapse on legacy integrations or edge-case data, so the real headcount savings never materialize and companies quietly keep paying for the human oversight layer they thought they could cut. It won’t be that way forever though

[Janakiram MSV @janakiramm]

SKILL.md is eating MCP Servers, and that's a good thing

Your MCP servers are burning 50,000 tokens just to teach an agent what a 200-token markdown file already knows.

Brad Feld runs an entire company on 12 skill files. No app. No workflow engine. Just markdown in a git repo.

Sentry's David Cramer says it bluntly. Many MCP servers shouldn't exist.
The problem? Teams keep building MCP servers for knowledge problems. But

MCP was designed for execution problems. The difference is costing you 50x in wasted context and worse agent reasoning.

I wrote the decision framework for getting this right.

What's your skill-to-MCP ratio looking like?

https://t.co/7xC39JPUjB

[aleatoricus @aleatoricus]

AI Twitter's business model is rediscovering the same five problems every six months (as AI evolves). For likes.

[Ethan Mollick @emollick]

Every damn day, another post with a thousand plus likes for a year old "breaking" paper that should "scare everyone using AI" because of issues with "latest top models" like Llama 4 and o3.

(The paper was good & multi-turn is hard, but, again, big progress since it was written.)

[aleatoricus @aleatoricus]

@bcherny @mritunjay394 good thread. In my case I'd reframe as working in chat "projects", but for Claude Code. Sharing context is a huge win @bcherny

[aleatoricus @aleatoricus]

New Anthropic study shows young workers are less likely to be hired into AI-exposed occupations. AI isn't taking jobs *yet*. What it's doing is quieter: companies just aren't filling certain roles anymore.

https://t.co/aKcgpbBG3x

[Claude @claudeai]

New in Claude Code: Remote Control.

Kick off a task in your terminal and pick it up from your phone while you take a walk or join a meeting.

Claude keeps running on your machine, and you can control the session from the Claude app or https://t.co/er6Blrr63e

[aleatoricus @aleatoricus]

@airtransat "503 Internal Error" is a system error, caused by a bug or infrastructure issue. Please forward this message as soon as possible to your IT department.

[aleatoricus @aleatoricus]

hey @airtransat any plans to make the seat selection work again? It's been almost two days with a 503 Internal Server Error (service's down). Tried via phone and on top of being 2 hours wait, the call drops.

[aleatoricus @aleatoricus]

@denny_zhou

[aleatoricus @aleatoricus]

@WatcherGuru Someone found their old HDD.

[aleatoricus @aleatoricus]

@PearsallRo77487 @RealDiamondNuts @elonmusk @Tesla I’ve test driven a Tesla model 3 (FSD) and an Equinox (Super cruise). I ended up buying the Tesla and I do not regret in terms of semi-autonomous driving. Also, one can only use super cruise on highways last time I checked, whereas FSD is for all roads. Oranges and apples.

[aleatoricus @aleatoricus]

@rodrigodasilva O tapa no Chris Rock virou coisa de esquerda/direita também?

[aleatoricus @aleatoricus]

@Cardoso @Gaa14Gabriel Tecnologia de cancelamento avançou bastante né

[aleatoricus @aleatoricus]

@benjamincowen Your floor’s bleeding against Ethereum