# Ben Shapiro > Applied AI, Product Builder, Mindful Technologist. I help teams decide what to hand to AI, and how to know it worked. From December 2024 to September 2026 I was co-founder and Head of Product & AI Strategy at Cadre AI, an AI consulting firm and Anthropic partner. Before that: Director of AI at Tide Rock, a private equity firm; three years on the founding team of an electric shuttle and school bus maker; and a startup that taught seniors to use technology. Experience: 50+ organizations, 20+ industries, 65+ AI trainings and talks. Organizations I worked with are described by sector and size, never by name. Every Markdown link below is a plain-text copy of the page. ## About - [About](https://shapyro.com/about): the longer story, from the stage to applied AI. - [Résumé](https://shapyro.com/resume.md): one page, plain text. PDF: https://shapyro.com/ben-shapiro-resume.pdf - [Contact](https://shapyro.com/contact): hello@shapyro.com - Elsewhere: [LinkedIn](https://www.linkedin.com/in/benshapyro/), [GitHub](https://github.com/benshapyro), [X](https://x.com/benshapyro) ## Work - [An Eval for Coding Agents, and the Three Bugs I Found in It](https://shapyro.com/work/gauntlet-agent-eval-harness.md): Own build. A harness that gives a model a coding task, grades the fix against hidden tests in a sealed container, and reports whether the model solves it on every try, not just one. The most useful thing it produced was a list of three ways it was wrong. - [Choosing an AI Platform for an Investment Office](https://shapyro.com/work/investment-office-ai-platform.md): Investment office of a large institutional investor. An investment office with a small tech team had to pick an AI platform and defend the choice. I turned its use cases into workflows a vendor could be tested on, held six vendors to the same seven gates, and laid out two paths. The executives chose the one our deck didn't lead with. - [Knowledge an Agent Can Act On](https://shapyro.com/work/context-layer): Own build. I rebuilt my personal context layer so AI can do real work from it without me checking every line. Then I designed Canon, a context layer for a whole company. - [A Personalized Claude Coach and the Meta-Skill That Builds It](https://shapyro.com/work/claude-coach-and-configurator.md): Own build. A Claude coach built for each firm we trained, so help was still there after the session ended. Then a meta-skill that builds the coach. The test was a colleague setting one up with no help from me. - [Rebuilding a Financial-Aid Estimator From a Spec With Claude](https://shapyro.com/work/financial-aid-estimator-rebuild.md): Private career university. Led the strategy and scoped the tool. Later I rebuilt it from a spec with Claude, in about an hour and a half of my own time, to test my spec method. - [Finding the Real Bottleneck at a Federal Supply Contractor](https://shapyro.com/work/federal-supply-contractor-ai-roadmap.md): Federal supply contractor with two operating companies. A holding company with two federal supply businesses asked what to do with AI, and in what order. I ran the discovery, wrote up the use cases and priced a 12-month program. The most useful finding was about lost quotes, and its first fixes needed no AI. - [Designing a Confidential Claude Setup for a Tax-Litigation Firm](https://shapyro.com/work/law-firm-confidential-claude-setup.md): Tax-controversy law firm. The founder of a tax-litigation firm had built his own Claude system, and only he could run it. The job was making it work for the whole firm under tax and bar confidentiality rules. A research brief changed the plan before anything got built. - [The Prompt and Skill Libraries I've Built Since 2025](https://shapyro.com/work/claude-skills-and-plugins.md): Own build. Eight libraries of prompts and Claude skills, from one long document of prompts to a plugin a team could install, a company marketplace and a public one. An overview of what each held and how I kept them in order. - [A Prompt Library I Shipped but Never Launched](https://shapyro.com/work/folio-prompt-and-skill-library.md): Own build. One place for our team to find a prompt or skill, fill in the blanks and copy it. I built it with Claude Code in a month. Nobody used it. ## Writing - [The Factory Floor](https://shapyro.com/writing/factory-floor.md): 2026-11-14. Follow one order down a production line staffed by AI agents, then turn the dial and watch where the people go. A scroll-through explainer of the software factory. - [Robot in a Room](https://shapyro.com/writing/robot-in-a-room.md): 2026-11-10. One robot, one sealed room, fourteen steps. A scroll-through explainer of how a chatbot becomes an agent. - [The Great Inversion](https://shapyro.com/writing/the-great-inversion.md): 2026-03-15. Your org chart was designed for expensive execution. That execution is getting cheaper every month, and the org chart has to change with it. - [The Organizational 95%](https://shapyro.com/writing/the-organizational-95.md): 2026-03-13. Most AI pilots never pay off. The gap isn't technology. It's the organization. - [The Death of Friction](https://shapyro.com/writing/the-death-of-friction.md): 2026-02-06. When friction disappears, every idea gets tested. That changes who wins. ## Optional - [RSS feed, full text](https://shapyro.com/feed.xml) - [Sitemap](https://shapyro.com/sitemap.xml)