Multiple AI systems.
One human
judgment loop.
“Used AI” is too broad to describe the Phase 2 process. The useful review question is how problems moved from exploration to comparison, human selection, architecture, implementation or documentation, and finally into a mapped portfolio under one accountable decision-maker.
The path is part of the review surface.
It is not the verdict.
The formation path matters because it exposes how problems were framed, alternatives were explored, decisions were selected and outputs were externalized. That process is reviewable, but it does not by itself determine novelty, market value or independent validity.
Human-directed
The founder determined priorities, sequencing, acceptance criteria and which ideas survived.
Multi-model
Different frontier AI systems were used for exploration, refinement, comparison and pressure-testing rather than treated as one authoritative source.
Iterative
Ideas could be reframed, rejected, split, recombined or abandoned before they became architecture, documentation or implementation work.
Externalized
Useful reasoning was turned into reviewable artifacts: specifications, diagrams, pages, implementation records, tests or mapped research directions.
From question to mapped asset.
The new Phase 2 overview already summarizes the method. This page expands it into an explicit review sequence without pretending every project followed an identical script.
Frame the problem
Define the actual problem, constraint, target user/system and what would count as a useful result before asking models to generate solutions.
Explore in parallel
Use multiple frontier AI systems to expand the search space, enter unfamiliar domains, surface alternatives and expose different reasoning paths.
Compare & pressure-test
Challenge assumptions, compare framings, look for contradictions and use cross-model disagreement as a reason to inspect—not as a vote that determines truth.
Select & synthesize
The human judgment loop decides what is coherent, what connects to the broader architecture, what must be rejected and what deserves further work.
Externalize the system
Convert surviving reasoning into a specification, architecture, research note, product concept, review document or other inspectable artifact.
Implement / test where applicable
For selected technical assets, move beyond documentation into code, operational logic, internal tests or benchmarks. This step is asset-specific, not universal.
Map maturity & route review
Classify the output by what actually exists—research/architecture, implementation, internal testing or independent review—and route deeper questions to the relevant page.
Re-enter when the answer is weak
Wrong turns, contradictions and discarded approaches can trigger another exploration cycle. Iteration is part of the formation method, not evidence that the final claim is automatically correct.
AI expands the search space.
Human judgment owns the decisions.
AI systems were used for
- Exploration and alternative generation
- Drafting and restructuring
- Cross-model comparison
- Technical explanation and hypothesis formation
- Critique, contradiction search and stress testing
- Assistance with documentation and implementation work where applicable
The human remained responsible for
- Problem selection and priorities
- Choosing which model output to trust, reject or investigate
- Cross-domain synthesis and architecture coherence
- Acceptance criteria and maturity labeling
- What becomes public, restricted or reserved
- Ownership, accountability and downstream decisions
One person can touch many role surfaces.
That does not create many credentials.
AI can compress work that normally spans several organizational functions into one founder-directed operating loop. The relevant question is whether that work is externalized and reviewable—not whether one person can claim the professional credentials of every role touched.
Domain entry & synthesis
Learning, comparison, theory formation and structured investigation across technical and product questions.
Problem & system framing
Turning observed gaps into product concepts, architectures, user flows and capability maps.
Architecture & implementation
Design work plus code/testing for selected assets, while keeping asset maturity explicit.
Externalized reasoning
Converting working thought into specifications, pages, maps, review packages and structured records.
Self-critique & routing
Building questions, failure conditions and review pathways without treating self-review as independent validation.
Portfolio coherence
Choosing what connects, what remains separate, what should advance and what should be discarded or deferred.
Outputs have different forms
and different maturity.
The formation method can produce different kinds of outputs. The correct review standard is to identify what exists, classify its maturity, and inspect that output on its own terms.
Research / Architecture
Conceptual models, theories, system maps and design specifications.
Implementation
Code, operational logic or working technical artifacts for selected systems.
Internal Testing
Internal experiments, benchmarks or controlled tests where available.
Review Architecture
Boundaries, hard questions, comparison frames and routing designed to make later diligence possible.
What this method can establish.
What it cannot establish.
What the method establishes
- Parallel / multi-model exploration under one human judgment loop
- Exploration → comparison → human selection → externalization
- A reviewable formation path rather than a purely private creative process
- Iteration, rejected alternatives and pressure-testing as part of the workflow
- Clear separation between AI assistance and human accountability
What the method does not establish
- That every asset reached implementation or testing
- That cross-model agreement is independent validation
- That documentation density proves novelty, patentability or market value
- That role compression replaces licensed or independent specialists
- That all technical or IP-sensitive detail belongs on the public site
- That the resulting portfolio automatically satisfies the OPU value thesis
After method,
test provenance and inspect the work.
The method explains how Phase 2 work was formed. OPU review then reconstructs provenance, while technical review inspects individual assets, maturity states and validation requirements.