# Cordanis Research

Original definitions, comparisons, and operating notes for teams applying governed AI to complex B2B sales.

Last updated August 16, 2026.

## Articles

- [What AI should prepare and what it should never execute alone](/research/what-ai-should-prepare-vs-execute): A practical boundary between scalable preparation and consequential action that remains under specific human authority.
- [Why human approval is a system boundary, not a button](/research/human-approval-is-a-system-boundary): Five tests for meaningful approval, from exact scope and reviewed-content integrity through execution and audit evidence.
- [What institutional memory means in complex sales](/research/institutional-memory-in-complex-sales): A definition of shared account memory that preserves sources, decisions, relationships, corrections, freshness, and privacy across handoffs.
- [The named-account operating model for sales AI](/research/named-account-operating-model-for-sales-ai): How AI should allocate attention, preserve context, coordinate teams, and measure coverage when the account book is finite.
- [How to evaluate an AI sales system](/research/how-to-evaluate-an-ai-sales-system): Eight connected tests that follow one real workflow from context and evidence through provider truth, audit, cost, and failure behavior.
- [What productive account coverage means](/research/productive-account-coverage): A definition of coverage that combines honest denominators, buying-role access, freshness, justified work, yield, cost, and reasons to stop.
- [What is an AI sales harness?](/research/ai-sales-harness): A definition of the governed layer between AI agents and a sales team's real book of business.
- [AI SDR vs. sales copilot vs. sales harness](/research/ai-sdr-vs-sales-copilot-vs-sales-harness): An honest comparison of the three ways to put AI on pipeline generation.