Resources

Field Research & Guides

Original work. No vendor funding. No AI-tool affiliations.

Guide

The No-Bullshit Guide to AI in Software Development

What AI can actually do in software development, what the research shows it cannot, and the organizational prerequisites most vendors never mention. A data-driven reality check on the "coding is solved" narrative.

What's Inside

  • Why "coding is solved" is true for some people and misleading for most
  • The three areas where AI delivers documented, measurable ROI today
  • The five organizational prerequisites your vendors will never mention
  • A four-tier readiness model to assess where your organization actually stands
Download PDF Guide
Executive Briefing

Does AI Actually Save Money on Software Development?

The truth about AI-assisted software development costs. Written for the people who sign the checks.

The verdict, up front: AI makes your smallest problem cheaper. It makes your biggest problem more expensive.

What's Inside

  • The lifecycle curve: Euphoria, Plateau, Decline, The Wall
  • Token economics and why autonomous agents have no cost ceiling
  • What Uber and Shopify actually did, and what made it work
  • Six independent research sources, cited and explained
19%
Slower
Experienced developers in a randomized controlled trial. They estimated 20% faster. (METR 2025 RCT)
45%
Critical flaws
Of AI code generation tasks introduced critical security flaws in controlled testing across 100+ models. (Veracode 2025)
10×
Duplication
More duplicated code blocks since AI coding became standard. Every duplicate is a future bug multiplier. (GitClear 2025, 211M lines)
Download the Briefing
Guide

The No Bullshit Guide to AI

A practical, executive-grade guide to generative, predictive, and agentic AI: what it is, what it is not, where it fits, and how to think about real risk and real value.

What's Inside

  • Clear definitions: generative vs predictive vs agentic
  • Where RAG actually fits (and when it does not)
  • Practical risk framing: data, security, governance, and liability
Download PDF Guide

Based on GitClear, METR, Veracode, DORA, McKinsey, and Stanford / Harvard labor market research. No vendor funding. No AI-tool affiliations.