LoomStack
PLAYBOOK

Agentic AI at organizational scale.

The operating model for coordinating AI-driven engineering. Ten modules. Decision frameworks, not buzzwords. Start anywhere.

10 modules·~115 min total·Updated June 2026·CTOs, VPs, Platform leads

CTO

Start with the systemic view: why the org isn't improving, where to invest, and what infrastructure you need.

VP OF ENGINEERING

Start with the delivery view: what's breaking in your review pipeline, how to govern at scale, what to measure.

PLATFORM ENGINEERING

Start with the build view: how specs anchor agent work, what the harness needs, and the architecture decision.

All modules

Read in order or start anywhere.

Foundation· 00

The Coordination Bottleneck

Why is AI making my engineers faster but my org isn't shipping better?

Outcome: Diagnose why individual speed doesn't become org delivery

Self-Assessment· 01

Where Your Org Stands Today

What stage are we at, and what should we do next?

Outcome: Score your org across 5 dimensions, get a stage-appropriate plan

Practice· 02

Spec-Driven Development

How do I ensure AI builds what we actually need?

Outcome: Why intent definition is the highest-leverage investment, not model selection

Coordination· 03

Agentic Engineering

How do I coordinate multiple agents and teams without killing the gains?

Outcome: Operating model canvas: team topology shift, agent fleet design, decision tiers

Verification· 04

Harness Engineering

How do I trust agent output without reviewing everything manually?

Outcome: Two-layer harness architecture, eval gates, governance RACI

Automation· 05

Loop Engineering

How do I move from individual prompting to automated, self-verifying workflows?

Outcome: Loop design template, failure-mode catalog, cost framework

Learning· 06

Compound Engineering

How do I make each AI-built feature make the next one easier?

Outcome: Codify checklist and org memory taxonomy

Infrastructure· 07

The AI Engineering Control Plane

Build, buy, or hybrid? What does the coordination infrastructure need to look like?

Outcome: Build-vs-buy decision framework, reference architecture, Stripe audit checklist

Measurement· 08

Metrics That Don't Lie

How do I measure whether this is actually working at the org level?

Outcome: AI-attributed metrics framework, 4-metric CTO dashboard

Action· 09

The 90-Day Plan

What do I do Monday morning?

Outcome: Stage-appropriate 30/60/90 plan across three org sizes

How these connect

Three disciplines, one operating model

Coordinate

Agentic Engineering

How you structure agent work, define decision tiers, and redesign team topology for AI-speed execution. Modules 00, 01, 03.

Deep dive
Constrain

Harness Engineering

How you verify agent output, enforce policy, and build the eval and governance infrastructure that makes autonomy safe. Modules 02, 04, 07.

Deep dive
Automate

Loop Engineering

How you move from individual prompting to orchestrated, self-verifying workflows, and make the whole system compound. Modules 05, 06, 08, 09.

Deep dive

Why this playbook exists

Engineering leaders are getting the same question from their boards right now: "We've spent 18 months adopting AI tools. Where's the ROI?" Individual productivity is up. Organizational delivery metrics are flat. There's a gap between those two things, and nobody has given you a clear framework for closing it.

This playbook doesn't sell you on a specific AI tool or vendor. It gives you the operating model: the coordination layer, the governance infrastructure, the measurement framework, and the action plan that converts individual AI velocity into organizational delivery.

The terminology will change again. "Agentic engineering" might not be what we call this in two years. The operating model won't. These problems (coordination overhead, governance gaps, verification debt, attribution failures) are structural. They don't go away when the tool names change.

Each module produces one artifact a leader can use immediately: a diagnostic, a self-assessment, a decision framework, an architecture template, a 90-day plan. No buying anything required.

Co-Build Program

From playbook to production

We work directly with engineering leaders who are making this transition now. You bring the real constraints and team context; we help you build the coordination layer around them.