How HAIA Came to Be
The practice and the frameworks that became the HAIA ecosystem date back to Factics in 2012. The name itself was an attempt to make communication more casual with AI after injury limited use of my right arm. A name for the thing I was speaking with: HAIA, Human Artificial Intelligence Assistant. It carried a few versions as I made my way through the frameworks: Human AI Intelligence Architecture, Human Augmented Intelligence Architecture, Human AI Augmentation, others. After settling into the ecosystem, I decided to keep it as it originated, Human Artificial Intelligence Assistant (HAIA).
There was no plan, only the practice of using LLM or AI to assist me, the human, to do things, be it marketing content, social media posts, work letters, kids’ school letters, correspondence with legal assistants and lawyers, everyday email response, even the Governing AI book.
As this was not an intentional practice to build anything, the reactive practitioner work of operationalizing tools to solve AI’s real problems became something I needed to backtrack on and structure. Each aspect was a response to an AI failure, a limitation, or a risk that surfaced in real use. I built prompts and frameworks to solve issues as I encountered them.
In summer 2025, I completed the Elements of AI course and the Ethics of AI course, offered and credited by the University of Helsinki. That led to the adoption of academic rigor in studying the thought leaders who shaped AI, from builders to ethicists to policy figures. I built a thought leader list to learn who had influenced AI and how. The study of their publications, paired with simulated AI reviews from their perspective, shaped the work that followed and became the basis for the book Minds That Bend the Machine.
This page stands alone, updated periodically to share the HAIA ecosystem and the frameworks that influence it, from AI Governance to Content and Social Tools.
How I Work
This page is the living reference for how I do my work: the method governance and standards behind my AI use, my policy work, and my research practices. My initial AI use was limited to ChatGPT, then Perplexity for sources, in 2025 I openly disclosed I was working in a five model Multi-AI use system that was in practice but was not named. The first published HAIA framework was The HAIA-RECCLIN Model draft in September 2025, and the method’s first full deployment in practice produced Governing AI: When Capability Exceeds Control. The original story and architecture live in the founding paper, HAIA: Human Artificial Intelligence Assistant, which stands as published. This page carries the current state. The purpose, in plain language, is in The Other AI: What “A Human-AI Collaboration” Should Look Like (PDF), the first-person account of why this ecosystem exists and how the method runs.
Everything runs under Checkpoint-Based Governance: a named human stays in control at defined checkpoints, nothing advances without human approval, and disagreement is preserved instead of smoothed away. Every work product carries a maturity status: a rough draft holds the first third of the governance work, a working paper has passed Cross AI Platform Review and independent Navigator review with human oversight and holds two thirds, and a final publication is fully CBG compliant with its SCOPE source documents complete. For every claim that carries weight, the primary source settles the question. The platform run surfaces what needs checking, the independent synthesizer narrows the field, and a human reading the original record certifies it.

In Use
Method Governance (the process for creating content)
- Factics — The foundation since 2012. A fact, a tactic, and a KPI set before any AI touches the work, so every claim carries its evidence and its purpose.
- Checkpoint-Based Governance (CBG) — Constitutional authority layer with binding human checkpoints.
- HAIA-RECCLIN (Researcher, Editor, Coder, Calculator, Liaison, Ideator, Navigator) — Seven-role reasoning and dispatch framework that makes the AI show its work: role, task, output, sources, conflicts, confidence, expiry, Factics, recommendation, and decision. Third Edition published March 2026.
- HAIA-CAIPR (Cross AI Platform Review) — The same work run across an odd number of independent platforms (3, 5, 7, 9, or 11) spanning American, French, Swiss, and Chinese lineages. Agreement is treated as a flag, and dissent is preserved. The run excludes the working AI, which stays out on purpose and serves as the independent synthesizer and fact checker of the run with the human.
Tools (the process for distributing content)
- HAIA-WOPPA (WordPress Optimization Publication Prompt Architecture) — WordPress publication packager producing paste-ready Support and Content documents from finalized work.
- HAIA-SMART (Social Media AI Rating Tool) — Six-pillar social content rating with Visual Intelligence Module, Multi-Platform Output Suite, AI Pattern Risk Detection, and Dissent Log.
- HAIA-CORE (Content Optimization Reader Evaluation) — Substance evaluation for blog and long-form content.
- HAIA-MOON (Multimedia Operational Outputs for NotebookLM) — Production prompts for cinematic video, deep dive audio, and infographic outputs, with steering and disclosure controls.
Record Keeping (how I prove the methods and the work)
- HAIA-CARCS (Compliance Accountability Record and Case Study) — Ten-section accountability record of what happened, who decided, and on what evidence.
- HAIA-SCOPE (Source Custody Observable Publication Evidence) — Per-citation source custody record documenting the human source audit that moves a working paper to a finished publication.
Evaluation (how I know I am cognitively developing and collaborating)
- HEQ (Human Enhancement Quotient) and AIS (Augmented Intelligence Score) — Measurement framework for cognitive amplification through human-AI collaboration, with the HAIA-HEQ operational prompt for measurement runs. Growth is measured, and so are diminished returns and decline.
Proposed, Not in Use
The infrastructure and policy layer, built or specified and published as part of the AI Provider Plurality work. None of it runs in the daily practice, and that distinction is the point: the practice proves the method, and the proposals show what the method looks like at enterprise and national scale.
- HAIA Agents — Audit-grade multi-AI collaboration architecture, EU regulatory compliance edition. Model 3, the human governor working manually, is the default in use today and produced every published case study. Models 1 and 2 are proposed automation. Formerly designated AEGIS.
- GOPEL (Governance Orchestrator Policy Enforcement Layer) — Non-cognitive governance layer that automates without thinking: seven operations, SHA-256 hash-chained audit trail, zero cognition by design. Reference code built, never deployed. Extensions: post-quantum cryptography and confidential processing.
- HAIA-Overwatch — Adaptive security shield wrapping GOPEL from an external trust boundary. Code built, never deployed.
- VAISA (Verified AI Inference Standards Act) — Proposed legislation for verified AI inference standards, submitted to the 119th Congress as part of the AI Provider Plurality Congressional Package.
GitHub Logs and Frameworks Medium AI Posts
#AIassisted using HAIA Ecosystem