---
title: 'In the News: November 11, 2026 (Evening)'
description: 'Two operators describe the software factories they run. André Staltz says make agents fail closed; David Cramer prices one task at $1.32.'
canonical_url: 'https://darkfactory.dev/news/2026-11-11-evening'
markdown_url: 'https://darkfactory.dev/news/2026-11-11-evening.md'
collection: news
date_published: '2026-11-11T19:10:00-05:00'
date_modified: '2026-11-11T19:10:00-05:00'
---

# In the News: November 11, 2026 (Evening)


Two people who run software factories published accounts of them this week. André Staltz lists what he learned, including agents that stop when their environment is broken and automation that scales up slowly. David Cramer of Sentry walks through one small task, gives its cost, and says the self-building factory is a myth.

## 1. André Staltz says to make agents fail closed and automate slowly

**[I gave a talk recently about lessons learned running a Software Factory](https://x.com/andrestaltz/status/2120253818889768968)** · André Staltz · X, November 10, 2026

Staltz posted a numbered summary of a talk he says he gave recently. We read the post in full. We did not find a recording or slides, so what follows is his own summary of his talk. He defines a software factory as a system of agents that build software with humans occasionally in the loop, and says the engineering craft moves into agent instructions and infrastructure. His good reasons to build one are less tedious work, less slop, and more trust in what agents build. His poor reasons include automating everything as soon as possible and wanting to stop looking at code.

On review he makes three claims. Work out what you check when you review a pull request, then teach an agent to check those things and report on them. He says he now trusts bot reviews more than his own, because they are systematic and can be audited later, while his own review is ad hoc. And he says people should write and read English, with agents following a set reporting process when they describe pull requests, because authoring in instructions while reviewing in code makes review the bottleneck.

Two lessons close the list. Narrow each agent with rules and linting and make it fail closed: an agent that lacks a correct environment should fail, and the person should fix the environment. The second: "Premature automation is the root of all evil." He advises scaling up little by little and monitoring.

Staltz's pinned post announces a release of Fabrica, a tool for running a software factory, so this is advice from someone with a product in the category. The post gives no measurements. It showed 184 likes, 292 bookmarks and 14.2K views when we read it on the evening of November 11, about 27 hours after posting.

**Why it matters:** Fail closed is a rule a team can write into a harness this week: a broken test environment stops the run, and the agent is told not to work around it. The review claim needs more caution. Reading an agent's English report in place of the diff works only if the report is accurate, and the post does not say how he checks that.

## 2. Sentry's David Cramer prices one factory task at $1.32

**[I have a Slopware Factory](https://x.com/zeeg/status/2120629042542194825)** · David Cramer, founder, Sentry · X Article, November 11, 2026

Cramer describes Junior, which he calls Sentry's open-source internal Slack agent, and says its name is on more commits than any other single contributor at the company. He walks through one request he says he made a few hours earlier: delete a kanban board experiment from Junior's own codebase, with no repository or other context supplied. He says Junior searched past conversations to fill in the context, did the work, and came back to Slack with a pull request. Cramer gave the code a quick glance, approved it, and let it merge automatically. He puts the total at $1.32 and five minutes of his time.

He is plain about the limits. He says the agent sometimes does well and sometimes produces garbage, that he approved without knowing whether every reference to the experiment was gone, and that CI would probably flush out any bugs. He calls this the state of the art for a company with real customers and liabilities, and adds: "Our system has memory - that doesn't make it self-improving."

This is one anecdote with one cost figure and no failure rate. The article showed 50 likes and 3,238 views under three hours after posting.

**Why it matters:** Cramer's account is a baseline for what a factory does today at a company with paying customers. A person states intent in chat, an agent finds the context and opens a pull request, a person approves, and CI is the check. The human review in his example was a glance, by his own description. A team copying the pattern should decide what CI has to prove before a glance is enough.