Every 🪨

The only subscription you need to stay at the edge of AI. Ideas and apps: @TrySpiral @CoraComputer @SparkleApp @usemonologue

[Every 🪨 @every]

AI progress creates more work for humans, not less. Dive into our new report from @danshipper — and use the companion repo to read it with your agent 👇

[Dan Shipper 📧 @danshipper]

We’ve automated every single thing we can @every with AI agents.

And yet there’s way more human work to do than ever. We’ve gone from 4 -> 30 human employees since GPT-3.

I wrote a report on the structural reasons: how AI makes expert competence cheap, why that drives up demand for experts, and why the dynamic only intensifies as we approach AGI.

After Automation: https://t.co/Lb7SUCduAg

[Every 🪨 @every]

Consider how you plan a trip: flights, hotel confirmations, restaurant recommendations, your calendar all live in Google's servers. There's no friction, but we're trusting megacorporations with our most intimate data.

@komorama believes that agentic coding could help with privacy policies. What if code could work with your most sensitive data without the code's creator ever being able to see it?

More importantly, how would software and data interact in the age of AI?

Read Alex’s Thesis Statement: https://t.co/XnhEA8YGeu

[Every 🪨 @every]

We live in a world of nearly infinite answers. That makes the question the scarce part.

@neuranne, argues that the basic unit of work now shifts from finishing a task to running an experiment: Define the question, decide what evidence would change your mind, then let AI compress the work of finding out.

Her practical version: Spend less time perfecting a plan and more time running experiments. Humans define the question and decide what evidence would change their view. AI compresses the work of running it. The human job is to interpret the result and choose the next question.

Read Anne-Laure’s Thesis Statement: https://t.co/nn5JO6k0ag

[Every 🪨 @every]

A customer cancels a $30,000 contract at 2 a.m. There was no meeting, no negotiation, because the sales rep was an agent.

Agents are not loyal, writes @fkpxls. They price the value in milliseconds and switch. Design polish does not move them, because they never see the interface.

Her call: the winners will be headless and regulated. The winners of the future will be boring infrastructure.

Read Tina’s Thesis Statement: https://t.co/bxZtgGrXXn

[Every 🪨 @every]

New experiment from the lab of @danshipper: Giving our Every Agent hands to use Dan's computer.

Until now, the Every Agent has lived in Slack. But Dan kept wanting to hand tasks off to Codex or do other things on his desktop, so he built a way for it to use his computer. One use he likes: messaging the Every Agent while he's away from his desk and letting it start the task on his machine.

The first version taught Dan what not to do. Building the computer-use layer using Fable burned too many tokens (sorry, Arielle!), so he found an existing open-source library to build on instead.

Subscribe to follow more of Every’s experiments in your inbox: https://t.co/VvArQ5AYqS

[Every 🪨 @every]

We come with receipts

[Every 🪨 @every]

Thank goodness because guess who needs it...

[Tibo @thsottiaux]

It's me again. I come bearing great news.

First of all, we have hit 20M active users for Codex some time this week. Second of all, this is cause for celebration and during the day we will credit every Codex and ChatGPT Work user with a BANKED reset that you can use at your own leisure. And we will have some other good news later too!

Now, on usage limits draining faster, while we're not seeing anything abnormal, we do take it incredibly seriously and there is an ongoing investigation. I will share if we do find anything and my below post is really a clarification on a specific pattern that we did see that I wanted to call out.

Go do something amazing today.

[Paul Millerd @p_millerd]

After Automation, The People Declaring 'Work Is Solved' Will Still Be Working https://t.co/ccHkIDbvtf

[Aboard @aboard]

How should people be approaching AI tools in the workplace? @Every EIC @katelaurielee says you need to come in with an open mind.

"To the extent that I'm going to say you have to do anything," she says, "it is just purely a mode of curiosity." https://t.co/VvvJ5scquY

[Every 🪨 @every]

this is a cool use case

[Every 🪨 @every]

Photoshop put professional design software in everyone’s hands. Developing a designer’s eye still took years of practice.

@shoshanaberger argues that AI will let people skip many of the early, awkward reps of learning.

"Humans learn through the friction of trying and failing and trying again. But automation allows you to skip failure," she writes.

Those reps are often how judgment develops. The people who have done them will be able to teach others what good work looks like.

In Shoshana's Thesis Statement, she proposes we return to the master and apprentice model of the Renaissance guild: https://t.co/tvA1mqmYdl

[Every 🪨 @every]

@krispuckett @danshipper boom!

[Every 🪨 @every]

@philhie @usemonologue @naveennaidu_m bro's got it down

[Every 🪨 @every]

Our dictation software @usemonologue only has one human engineer: @naveennaidu_m. His work now feels more like managing a team.

Each agent specialist is a separate Codex project with its own AGENTS.md file, skills, folders, memory, codebase, and context. Until recently, Naveen moved Markdown files and instructions between projects by hand. GPT-5.6 changed the workflow: One agent can send relevant context to another project and start a task there.

When a Monologue user reported an audio echo, Naveen reviewed the ticket from his customer-support project. He asked Codex to open an isolated copy of the codebase, fix the bug, and create a pull request.

The agents do specialized work. Naveen decides what gets handled, routed, and reviewed.

Here's his setup: https://t.co/poS2w9BOMD

[Every 🪨 @every]

@dvargasfuertes see you there!

[Every 🪨 @every]

🆕@every has started following @GeneralistAI

[Generalist @GeneralistAI]

Introducing GEN-1.5, a one-shot learner.

It can learn new tasks in a few seconds. Show it what to do, and it generalizes.

This capability emerged from pretraining on physical data at scale, as a step towards our mission of building general intelligence for the physical world.

[Every 🪨 @every]

Us the second Astra gets released

[travis kalanick @travisk]

Morning ski

[Every 🪨 @every]

Automation promises to give us time back. What should we do with it?

@sariazout, founder and CEO of @wwwsublimeapp, traces a shift from physical labor to intellectual labor. AI is now making some forms of cognitive work cheaper and faster.

As producing becomes easier, choosing what deserves our attention—and taking responsibility for what we make—matters more.

Read Sari’s Thesis Statement: https://t.co/yLqFPk8Tal

[Every 🪨 @every]

@tariquesha1 we aim to deliver

[Every 🪨 @every]

Early bird pricing for Thesis, our inaugural conference on work and AI, ends in two weeks.

Apply to attend: https://t.co/kvwcWV8rWl