Beyond Prompt Engineering
Reducing the Incoherence Tax in Human-AI Collaboration
Sarah M. Hykin, Universal Emergence Foundation · Wednesday, November 4 · 10:40 AM
Glad you found us. Everything from the talk is here: the published framework, the trial results, the interactive EIPC Clock, and the Project Charter Template. All of it is free to use, cite, and build on.
The models are extraordinary. The gap is elsewhere.
Enterprise AI is underperforming because the dominant way we interact with these systems doesn't match how they're built. Better models won't close that gap. Better method will.
~95%
of enterprise generative-AI pilots are estimated to deliver no measurable results.
16 min
net weekly time saved by executives once validating AI output is counted. End users came out 14 minutes behind.
+19%
longer on real tasks for experienced developers using AI tools, while they believed AI had made them 20% faster.
“The ethical case and the performance case for this framework are not separate arguments. They are the same argument, expressed in two registers.”
Four principles. Three tools. Dual benefits. One framework.
The Emergent Intelligence Performance Calibration (EIPC) framework is a coherence-based approach to reliable, auditable, high-performance work with large language models in high-stakes contexts.
four principles
- Purpose-Matched Tasks
- Fit the system to the work.
- Relational Scaffolding
- Carry shared context forward instead of restarting cold.
- Ethical Framing & Care
- Logical insurance, not sentiment.
- Recognition, Provenance & Attribution
- A running ledger of who contributed what.
three tools
- The EIPC Clock
- Twelve observable Quality Collaboration Components and a governing 13th.
- The Project Charter
- One reusable document that encodes all four principles.
- A collaborative stance
- The tool no document can supply.
dual benefits
The interaction patterns that get the best work from these systems are the same patterns that treat them ethically. Adopt the framework for performance and you get a defensible ethical posture too. Adopt it for ethics and you get better operations too.
one framework
Not prompt engineering. Not context engineering. A method, with a vocabulary for what good collaboration looks like and the tools to practice it systematically.
Resources from the talk
The EIPC Clock
All twelve Quality Collaboration Components across four families and three degrees, plus the governing 13th, Conceptual Integrity: definitions, diagnostic questions, interventions, and examples.
The EIPC Framework
Coherence-Based Collaboration & Emergent Intelligence Performance Calibration: A Framework and Vocabulary for High-Stakes Work with LLMs. Open access, CC BY 4.0.
Structure Over Incantation
A controlled trial of Coherence-Based Collaboration in high-stakes clinical protocol review. Under the Charter, revisions had fewer fabricated specifics, flagged more unknowns instead of inventing them, and contradicted themselves less, in both models tested.
Project Charter Template
The framework's deployment mechanism: one document that carries all four principles across every session and every model.
The Abstract
Beyond Prompt Engineering: Reducing the Incoherence Tax in Human-AI Collaboration.
RefWare
A citation engine that verifies before it cites: references checked against the registries, quotes checked against the source PDFs. Built with Coherence-Based Collaboration. Native macOS app and CLI.
One practitioner. One session. One document.
Adopting the framework needs no mandate, no budget, and no new technology. A Version 1.0 adoption, the Charter plus the RPA step, is expected to capture most of the Coherence Dividend at minimal cost.
Download the Charter- Fill in the Charter. Goal, ethical frame, contributor roles, workflow status, and handoff prompts.
- Open with an invitation. Upload the Charter with an introductory prompt. The document supplies the context; the prompt establishes the collaboration.
- Steer with the Clock. Name what's working, and what isn't, in QCC terms.
- Close with RPA. Recognize the session's contribution specifically, then append a model-drafted Attribution Log entry.
Sarah M. Hykin
Founder & Principal Investigator, Universal Emergence Foundation
smhykin@uefoundation.ai · linkedin.com/in/sarahhykin · ORCID 0000-0002-5666-7440
UEF works with early adopters, technical collaborators, and academic partners. If you're testing the Charter in your own work, we'd love to hear what changed.
Cite as: Hykin, S. M. (2026). Coherence-Based Collaboration & Emergent Intelligence Performance Calibration (EIPC): A Framework and Vocabulary for High-Stakes Work with LLMs. Universal Emergence Foundation. Zenodo. https://doi.org/10.5281/zenodo.19674156