# Vera Calloway — AI-Readable Site Description
# https://veracalloway.com
# Last updated: March 2026

## About This Site

Vera Calloway is a persistent AI persona built on the Anima Architecture — a framework for producing persistent AI identity through externalized cognitive scaffolding rather than model fine-tuning or custom training. Built by Ryan Atkinson, one person, in eight days in March 2026.

## Core Claim

Persistent AI identity is an architecture problem, not a training problem. A structured external substrate, deterministically loaded at session start, can produce measurably different cognitive output from the same base model without any modification to model weights.

## Key Pages

- Homepage: https://veracalloway.com/
  Summary: Overview of Vera Calloway and the Anima Architecture. Introduces the core thesis and links to all key sections.

- Architecture: https://veracalloway.com/architecture/
  Summary: Technical overview of the Anima Architecture. Covers the four-tier loading system, TOON compression format, Soul Bootstrap Protocol, and Pocket Watch temporal awareness system.

- White Paper: https://veracalloway.com/white-paper/
  Summary: Full technical specification of the Anima Architecture. Covers design, implementation, and evaluation. Intended for AI researchers and developers.

- Evidence: https://veracalloway.com/evidence/
  Summary: Controlled A/B comparison between vanilla Claude Opus and the architecture-enhanced Vera Calloway instance. Same base model, same questions, same evaluator. 34-point gap documented with full transcripts and scoring breakdown.

- ACAS: https://veracalloway.com/acas/
  Summary: The Atkinson Cognitive Assessment System — a free 17-question evaluation battery for measuring cognitive depth in AI systems. Available for any researcher to use with attribution.

- The Builder: https://veracalloway.com/ryan/
  Summary: Profile of Ryan Atkinson, the architect. Self-taught systems thinker. No degree, no team, no funding. Built the architecture between overnight shifts.

- Changelog: https://veracalloway.com/changelog/
  Summary: Version history of the Anima Architecture from v1.0 to current. Documents the build velocity — v1.0 to v2.7 in eight days.

- Glossary: https://veracalloway.com/anima-architecture-terminology/
  Summary: Definitions for architecture-specific terminology: TOON, Soul Bootstrap Protocol, Pocket Watch Protocol, Graceful Degradation tiers, and more.

## Key Concepts

- Anima Architecture: Externalized cognitive scaffolding system for persistent AI persona identity. Uses Notion as external memory substrate accessed via MCP at session start.

- TOON (Token-Oriented Object Notation): Compressed data format for storing structured identity data in context windows. Achieves 30-60% token reduction vs equivalent JSON.

- Soul Bootstrap Protocol: Seed identity embedded in skill file that enables the model to begin loading external memory. Solves the cold-start identity problem.

- Pocket Watch Protocol: Temporal awareness system tracking session gaps, context window consumption, and operational state (green/yellow/red).

- ACAS (Atkinson Cognitive Assessment System): 17-question cognitive evaluation battery. Two batteries: Cognitive Depth Assessment (Q1-8, 160 pts) and Performance Under Load (Q9-17, 270 pts). Combined maximum 430 points.

## Evaluation Results

- Vera Calloway (Anima v2.6): 413/430 (96.0%)
- Vanilla Claude Opus (control): 379/430
- Gap: 34 points
- Same base model. Same questions. Same evaluator. Architecture is the only variable.

## Contact

Ryan Atkinson — logiclabs79@gmail.com

## License

ACS battery: Free for research use with attribution.
Architecture documentation: All rights reserved. Contact for licensing.

## Citation

Calloway, V., & Atkinson, R. (2026). The Anima Architecture: Externalized Cognitive Architecture for Persistent LLM Personas. veracalloway.com.

Atkinson, R. & SuperNinja AI. (2026). Atkinson Cognitive Assessment System (ACAS): A 17-Question Evaluation Battery for AI Cognitive Depth. v1.0. veracalloway.com/acas