Research Studies

The evidence behind
the recommendations.

Where our Insights answer a question, a Research Study defends a finding. Each one takes a position on how AI should actually be deployed and backs it with cited evidence. Evidence over hype.

Published7 studies

Market Analysis

The Four Agentic AI Powerhouses, and What They Actually Do for a Business

What Perplexity Computer, Claude Cowork, OpenAI Codex, and xAI Grok each do well, where each one breaks, where the hours actually come back by industry, and why buying the subscription is the easy part.

16 min read/10 cited sources

Legal Research

How Public AI Tools Waive Attorney-Client Privilege

Why pasting client documents into public AI models can destroy attorney-client privilege, what recent courts have held, and how sovereign deployment preserves protection.

14 min read/7 cited sources

AI Research

LLM Social Reasoning Is Brittle: Why You Should Scope AI by Verifiability, Not Capability

LLM performance on social-inference benchmarks collapses under small perturbations. This study defends the position that AI work should be scoped by verifiability rather than by apparent capability.

13 min read/6 cited sources

Infrastructure Study

Stop Paying a Stranger by the Page to Read Your Own Files

Why firms whose entire liability model depends on document confidentiality should own their search infrastructure outright, what the architecture looks like, and the 36-month cost comparison against metered alternatives.

10 min read/9 cited sources

Infrastructure Study

Your Floor Does Not Wait for a Data Center to Answer

Why facilities running real-time equipment should process sensor and vision data on local hardware, how the architecture differs from cloud analytics, and the economics of owned versus metered inference at volume.

10 min read/8 cited sources

Infrastructure Study

Your Busiest Month Should Not Be Your Biggest Bill

Why operations handling high support volume should own their service infrastructure, how the architecture absorbs the repetitive queue without replacing the team, and the cost comparison against metered alternatives.

10 min read/7 cited sources

Infrastructure Study

Your Whole Schedule Lives in One Person's Head

Why field service and logistics operations should encode dispatch knowledge into a system that reads their live databases, how it handles the 9am schedule collapse, and the economics of owned versus metered optimization.

10 min read/6 cited sources

In ProductionComing Soon

Benchmark

Mid-Market AI ROI Benchmark

What owners actually recover in hours and dollars in the first 12 months of an AI infrastructure build.

Field Study

Automation Impact Report

Before-and-after operational metrics from real engagements across professional services, healthcare, and hospitality.

The Insights Library documents the frameworks and cites the external research these studies build on.

Want to see the numbers for your operation?

An Operational Snapshot benchmarks where your business actually leaks time and money, and what AI infrastructure would recover.

The evidence behindthe recommendations.

The Four Agentic AI Powerhouses, and What They Actually Do for a Business

How Public AI Tools Waive Attorney-Client Privilege

LLM Social Reasoning Is Brittle: Why You Should Scope AI by Verifiability, Not Capability

Stop Paying a Stranger by the Page to Read Your Own Files

Your Floor Does Not Wait for a Data Center to Answer

Your Busiest Month Should Not Be Your Biggest Bill

Your Whole Schedule Lives in One Person's Head

Mid-Market AI ROI Benchmark

Automation Impact Report

Want to see the numbers for your operation?

The evidence behind
the recommendations.