---
type: digest
source_type: x
category: reference
domain: enterprise-adoption
status: captured
captured: 2026-06-02
published: 2026-06-02
confidence: medium
tags:
  - alex-lieberman
  - businessbarista
  - software-factory
  - code-factory
  - agentic-ai
  - enterprise-adoption
  - workflow-migration
  - verifiability
source: https://x.com/businessbarista/status/2061573150719021542
raw_source:
  - "[[00 Raw/X Posts/2026-06-02 Alex Lieberman - Software Factory Ladder]]"
---

# Alex Lieberman - Software Factory Ladder - Digest

## Core claim

Alex Lieberman's June 2 X Article reframes the current "software factory" conversation for non-technical leaders: the goal is not engineers writing code faster with AI, but a production line where agents increasingly write, review, test, ship, monitor, and triage while humans design the line, set guardrails, and handle exceptions.

## Useful KB signal

The post is useful because it converts agentic software engineering into an executive diagnostic ladder:

- **Level 0:** human craft process
- **Level 1:** AI-assisted human coding
- **Level 2:** agents own tickets and open PRs, with human review
- **Level 3:** low-risk work flows through agent monitoring, triage, coding, review, tests, and auto-merge under guardrails
- **Level 4:** routine bugs are detected, fixed, reviewed, tested, and shipped before the team sees them

The strongest KB takeaway is that the maturity question is not percentage of AI-written code. The better question is how much of the delivery line runs without human touch, under which rules, with which evidence, and with what rollback path.

## Strategic interpretation

This strengthens the workflow-migration thesis. Software engineering is the first strong proof surface because it already has the ingredients agents need: tickets, code artifacts, tests, CI, PR review, production monitoring, version control, and rollback. Those ingredients make output verifiable and reversible enough to move humans from doing every step toward supervising the line.

It also sharpens the control-surface thesis. A software factory is a governed work system. The durable surfaces are:

- risk classification for auto-merge versus escalation
- automated tests the organization actually trusts
- agent-to-agent review
- production monitoring and triage
- secure access to code/docs/systems
- internal platforms agents plug into
- audit trails, rollback, and exception handling
- review capacity for higher-risk work

The "next factory" criterion is verifiability. Engineering, QA, analytics, data pipelines, and DevOps are naturally early because success signals are explicit. Finance/accounting reconciliation, support resolution, and operational sales/marketing follow where outputs are digital, repeated, measurable, and reversible. Strategy, design taste, novel architecture, executive judgment, and relationship work remain harder because correctness is delayed, subjective, or socially mediated.

## Caveats

- Treat the cited adoption metrics from Anthropic, Google, OpenAI, Ryan Carson, and StrongDM as source claims until paired with primary evidence.
- The post is also category positioning for Tenex Labs, which Lieberman says helps enterprises build software factories.
- Level 3 and Level 4 should be judged by actual operating evidence, not self-description: auto-merge rate, incident rate, rollback time, test quality, risk-policy coverage, auditability, and review bottlenecks.

## Source trail

- Raw source: [[00 Raw/X Posts/2026-06-02 Alex Lieberman - Software Factory Ladder]]
- X source: https://x.com/businessbarista/status/2061573150719021542

## Related notes

- [[05 People/Alex Lieberman]]
- [[03 Research/Agentic AI/Agentic AI Overview]]
- [[03 Research/Enterprise Adoption/Enterprise Adoption Overview]]
- [[06 Theses/The Workflow Migration Thesis]]
- [[06 Theses/The Control Surface Thesis]]
- [[09 Implications/Migrate Workflows Not Seats]]
- [[09 Implications/Pair Governed Runtimes With Workflow Owners]]
