I study how sustainability and technological change affect company economics, financing and long-term value. HHFinAi is my independent research portfolio: investment cases, source-linked financial models and analyst-controlled workflows.
Research focus: financial materiality · capital allocation · credit and relative value · climate resilience · responsible AI · stewardship
No installation is needed to read the findings. Each case distinguishes evidence, assumptions, analysis and unresolved questions.
| Investment question | Finding or analytical contribution | Read the work | Status |
|---|---|---|---|
| Heidelberg Materials: what must Brevik CCS earn to matter? | A premium break-even and incremental cash-flow bridge test the economics; a transition investment does not automatically justify a company-wide valuation premium. | Investment memo · Results and sensitivities | Retrospective real-issuer case; 25 February 2026 evidence cutoff; WATCH |
| EIB EuGB 2037: is a credible label enough? | Credit, contractual evidence, sustainability claims and entry price are separate judgments; matched historical pricing remains a material gap. | Investment memo · Evidence and model guide | Historical real-bond case; 16 October 2025 evidence cutoff; WATCH |
| Microsoft: what must water-resilient AI infrastructure earn? | An illustrative incremental cooling model solves the additional operating-benefit hurdle instead of assuming that water savings establish a return. | Research case · Model and results | Issuer disclosure context plus an illustrative model; prepared 4 October 2026; not actual site economics |
| Microsoft: how should an investor assess responsible AI? | A source-linked assessment turns environmental, workforce, IP, privacy and accountability questions into six observable proposed milestones. | Engagement case · Milestone ledger | Proposed engagement programme; no meeting, request, issuer response or outcome claimed |
Investment judgment: read the conclusion, the counterfactual and the strongest contrary case. Ask which evidence would change the decision.
Financial discipline: inspect the model inputs, units, cash-flow boundaries, sensitivities and break-even conditions. Reported facts, management expectations and analyst assumptions are not interchangeable.
Research execution: follow the sources and run the deterministic calculations. The Microsoft example includes 13 case-specific unit tests and 48 engineered metadata checks. These are software/contract checks, not LLM accuracy, measured time savings or investment alpha.
Evidence → economic mechanism → financial model → investment judgment → stewardship or monitoring question.
The original HHFinAi work is the implementation: counterfactuals, financial bridges, evidence boundaries, review controls and research-to-engagement handoffs. Third-party frameworks are attributed rather than presented as original inventions. AI assistance is disclosed; publication is not a claim that every source or assumption has received independent human validation.
Research standards and AI-use boundaries · Research development and open questions
Financial materiality · Bond diligence · Stewardship · Climate research
The wider suite covers nature, transition plans, carbon accounting, sustainable mandates and social/impact finance. These are supporting workflows, not evidence of completed research in every domain. Browse the full capability catalogue.
Healthcare and life-sciences tools are maintained separately at hh-health-AI.
Independent research and educational work, not investment advice, regulatory certification, an audited performance record or an autonomous trading service. The local engines organize and validate research; a human or separately authorized AI host retrieves evidence and performs substantive analysis. Cases and demonstrations carry their own dates and limitations. Material unknowns remain visible.