L1 — the commoditised layer
Credentials · how I use them · what I have codified
The framework’s own test for L1 is automate — what is codified enough that anyone (or any AI) can run it. This is where my recognised credentials live. Most of it is not uniquely mine.
Credentials and what I take from each
Computer Science (CS) degree
Technical fundamentals I reach for most — how systems fit together, how data flows, where they fail.
Organisational Psychology
People-and-systems lens — team dynamics, stakeholder-resistance patterns, organisational reading.
Accounting
P&L, balance sheet, T-accounts. Reconciliation between systems is still one of the most useful lenses I own.
Project Management
PMP (PMI) and Prince2 — phase-gated delivery, dependency management, stakeholder cadence.
Service Management
ITIL — service design, operation, incident discipline.
Change Management
ADKAR / ProSci methodology — applied for years, not formally certified.
One thread that runs through all of this: the tools above are commoditised. The judgement of which one to reach for in which situation is not. Waterfall or iterative? Light-touch integration or full absorption? Where to bend the rules and where the cost of bending is too high? That judgement is the part of me that acts before I can explain why. It lives further down the page, in L3b.
Where I am augmented — at a glance
Taking the areas those credentials and disciplines actually point at, the question becomes: what is holding the knowledge now? The model? A retrieval set wired in? Or is the work still on me? Filled circles mean strong coverage; empty means open.
| Area | In the model | Retrieval | Action / note |
|---|---|---|---|
| SAP / ERP documentation | Continue building the corpus as engagements demand | ||
| Accounting policies(AU federal / state / local) | Federal layer active; state and local still expanding | ||
| Telecoms / ICT | Broad industry coverage in model; AU-specific carrier regulation not benchmarked | ||
| M&A integration and divestment | Generic frameworks in model; my personal playbook not yet extracted | ||
| Capital portfolio management | Economics / business are among the lower MMLU-Pro categories (~78–83%); no personal-domain corpus yet | ||
| Foreign investment review (FIRB) · transfer pricing · valuations | Rules are public — retrieval build is straightforward | ||
| Project management(PMP · Prince2) | Fully in model; capability build pending | ||
| Change management(ADKAR / ProSci) | Fully in model; no dedicated retrieval or capability yet | ||
| Financial accounting | MMLU-Pro Business / Accounting ~80–85% for frontier; capability not yet built |
full three-quarter half empty
Method — how these values were derived (April 2026)
Model axis — triangulated against external benchmarks where public data exists:
- Project management · Change management — MMLU-Pro Management + Professional Psychology, frontier models ~85–90%.
- Financial accounting · M&A · Capital portfolio — MMLU-Pro Business / Economics, frontier ~78–85%. Business is one of the more reasoning-heavy MMLU-Pro categories; Economics is among the lower scorers.
- FIRB · transfer pricing · valuations — LegalBench (Gemini 3.1 Pro ~87.4%, GPT-5.4 ~86.0%) with a downward adjustment for AU-specific regulatory detail not well-represented in training data.
- Telecoms / ICT · SAP / ERP · AU accounting policies — no dedicated public benchmark; values are informed inferences from general-domain coverage patterns.
Retrieval axis — direct inspection of what is wired into my working context: documentation corpus, ingest pipelines, and Research Analyst capability state.
Frontier reference models (April 2026): Claude Opus 4.7, Gemini 3.1 Pro Preview, GPT-5.4. Overall MMLU-Pro: Gemini 3.1 Pro ~91.0%, Claude Opus 4.7 ~89.9%.