Jess Martin

aicoding.leaflet.pubCode Was Never the Asset - The Phoenix ArchitectureWhy AI makes the hidden economics of software unavoidable

1 May

Feeling this every single day. Trying to restructure my practices and habits to go deep on comprehension and understanding. aicoding.leaflet.pub/3maqpvianlc2...

@eugenevinitsky.bsky.social

12 Mar

Coolest thing I’ve read all year

Eugene Vinitsky 🍒@eugenevinitsky.bsky.social

www.percepta.ai/blog/can-llm... As a research lark at Percepta, Christos embedded a computer into an LLM, showed that it could solve the hardest Sudokus, and then as a side bonus built an exponentially faster attention

Jess Martin reposted

Eugene Vinitsky 🍒

irc.freeq.atfreeqIRC for the open web. AT Protocol identity. Pronounced like 'freak'.

16 Feb

the context is the software the code is just a cache

10 Feb

I wonder if GitHub is ready for 100-1000x in volume

10 Feb

every developer a 10x developer

6 Feb

Overnight on lifebuild.me I asked Codex to: - Swap out my in-product agentic loop for pi.dev - Refactor the entire UI, removing dead code - Create custom project icons in-app w/ Nano Banana Pro - Create a LifeBuild MCP server 📈 models ➡️ 📈 ambition

5 Feb

Something people don't talk about enough when working with AI is just how much *fun* it is. It's not just about productivity! Claude Code is fun. Conductor is fun. ClawdBot is fun. Lean into the joy, folks!

lifebuild.meLifeBuild — Reclaim Your SovereigntyLifeBuild helps you design, track, and achieve the life you want to live. Take back control of your time, energy, and priorities.

4 Feb

First batch of alpha invites going out for lifebuild.me Nervous, but excited! Feels so janky right now and there's a million things I want to fix, but really important to just get it out there.

Jess Martin reposted

Jake Lazaroff

@jakelazaroff.com

4 Feb

Replying to @joodaloop.com

i mean i'd love to see more efforts like OCIF but failing that i think people could start cramming their app's data into HTML, PDF, etc. wouldn't make interop seamless but it would be easier to read and adversarially edit or script around

canvasprotocol.orgOpen Canvas Working Group | OCWGWorking group for establishing interoperability between different infinite canvas tools.

So far, I've written about the Software Factory and the Software Library. Factory has gotten all the play. No one knows what to do with Library (or why they even need one). People saying things like "I run 3-4 agents at a time, life is good! we've arrived!" That's not the factory. Not even close.

We're talking hundreds of simultaneous agents. For that, you need context to maintain alignment. And that takes a Library. I'm not saying we the best solution yet, but I am saying this is going to be a MAJOR problem that more people need to start thinking about.

sociotechnica.org/notebook/con...

sociotechnica.orgThe Context Library of Alexandria (Part 1)Why agents make poor micro-decisions, and a five-dimension framework for delivering the context they're missing.

What will you do with your one wild l̶i̶f̶e̶ LLM-enhanced senior eng sw dev skillz?

The easiest path to malleable software: put an agent inside your app that has access to the repo for the product and can implement fixes and push PRs in response to the user pointing out problems. "Self-healing software"?

The Software Factory solves an operational problem: how do we orchestrate agents to deliver high-quality software. Once you know what to build, the factory gets to work. It doesn't solve the alignment problem. How do you keep the factory fed? Let me explain. x.com/jessmartin/...

And we're building an AI librarian to maintain it. His name is Conan. Conan watches context streams, decomposes into atomic notes, establishes links, assembles slugs on demand. His human boss makes judgment calls. Conan handles volume.

sociotechnica.orgThe Context Library of Alexandria (Part 1)Why agents make poor micro-decisions, and a five-dimension framework for delivering the context they're missing.

Full initial writeup with the five-dimension framework: sociotechnica.org/notebook/co... PS: If you're building in this space, I want to compare notes.

Give an agent the full context and the micro-decisions improve. Not because the model got smarter—because the context got richer. This is a management problem, not a model problem.

We're calling this a Context Library. There are two related context: - The Library: complete knowledge base covering past, present, future - The Context Constellation: curated context assembled for a specific task

None of this gets written down. Almost none of it reaches the agent. Agents typically receive: - WHAT: specs, requirements - HOW: codebase, process instructions

What they don't get: - WHERE: What's adjacent? What breaks if you touch it wrong? - WHEN: What did we try before? Why did it fail? - WHY: What objective does this serve? WHERE, WHEN, and WHY are the missing bundle.

There's a growing body of work on why separate AI agents struggle to make aligned decisions. They miss nuance. Break adjacent systems. Repeat past mistakes. The usual diagnosis is the agents aren’t good enough yet. But I think that's wrong.

The real issue is we've been starving agents of context human devs absorb by osmosis: - Hallway conversations about why the last approach failed - The senior engineer who says "we tried that in 2024" - Ambient awareness of what's adjacent, what's fragile

29 Jan

I was wrong. It's not JUST the year of the Software Factory. It's the year of agent orchestration, generally. Across all domains. I wasn't seeing the larger movement. Now I can't un-see it.

conductor.buildConductor - Run a team of coding agents on your MacCreate parallel Codex + Claude Code agents in isolated workspaces. See at a glance what they're working on, then review and merge their changes.

I use conductor.build every day and love it. I wonder if they are thinking about higher-level orchestration? I hope so! someone needs to to build it.

Jess Martin@jessmart.in

Here’s the shift I imagine: tasks flow through a factory line. Imagine software dev as a literal factory. Each stage a station. Tasks enter and flow through, station by station, and emerge as deployed software. Developers don't shepherd. They build and operate the line.

maggieappleton.comGas Town’s Agent Patterns, Design Bottlenecks, and Vibecoding at ScaleOn agent orchestration patterns, why design and critical thinking are the new bottlenecks, and whether we should let go of looking at code

2026 is the year we will stop hand-writing code and fully mechanize software engineering. This might sound like hype. And certainly there’s a lot of hype going around! But something substantial lies underneath the bluster. Here’s how I’m thinking about it.

If you want to catch up on the latest hot projects in this space, Yegge's Gastown is quite interesting. Check out @maggieappleton.com's breakdown of Gastown to clean the good stuff: maggieappleton.com/gastown

*glean

How might this change software architectures? You should be looking for things like @chadfowler.com's Phoenix Architecture: aicoding.leaflet.pub/

aicoding.leaflet.pubThe Phoenix ArchitectureGenerative AI coding demands what we've always known: modularity, clear boundaries, disposable components. Principles that scaled human teams are now table stakes. Here, we make the implicit explicit

It’s not just theory, fascinating as that is! We're trying out the factory approach while we build an app called LifeBuild. lifebuild.me

Props to @schickling.dev—he sent me down this path last July with this tweet: x.com/schickling/...

We don't know the limits yet. Individual devs using AI are already seeing 5-10x. A factory line? Maybe 100x? The next Google might be 10 people operating a finely-tuned factory.

sociotechnica.orgThe Year of the Software Factory

I wrote this up here: sociotechnica.org/notebook/so...

Why does this matter? In a factory, improvements compound to the whole org. Optimize one station and every task that flows through benefits. Forever. This is why manufacturing ate craft production. Not skill—system. And hand-crafted software will get eaten the same way.

For 50 years, developers have shepherded tasks through the software development lifecycle. You pick up a ticket, carry it from planning to implementation to review to deploy. I think that model is ending.

25 Jan

Liquid Glass. Love the Liquid. Loathe the Glass.

24 Jan

Freeform on iPad Pro 12" is pretty nice, actually.

19 Jan

Spending the day working on automating my software factory. Becoming more and more convinced that manufacturing is an appropriate metaphor and mechanism for managing software production.

Claude Code makes it so easy to go from "bug report" or "feature request" to open PR, it's kinda wild.

Just had the wild realization that since I have agents *inside* my product, I can give them a tool to log feature requests + bugs and as they are working with the user. Magic!

discord.comDiscord - Group Chat That’s All Fun & GamesDiscord is great for playing games and chilling with friends, or even building a worldwide community. Customize your own space to talk, play, and hang out.

Hey folks, I'm "hosting" a @livestore.dev Community Hour tomorrow @ 11am ET in the LiveStore Discord: discord.com/events/11544... I'll be adding the new Sync Status API to my LiveStore app. Drop by, lurk, ask questions, or work on your own app!

No better time to build that software thing you've been kickin' around in the back of your mind!

14 Jan

Claude (code) operating my computer is like the most fun computing I've had... maybe ever?