Articles
Longer arguments on finance, governance, execution, and the organisational consequences of AI.
- (31) From Daily Reading to Cumulative Expertise
An agent can maintain the archive. The reader still has to reconstruct, challenge and apply what it contains.
- (30) Owning Shares Does Not Guarantee Alignment
A substantial executive shareholding can coexist with conflicting incentives. The revealing questions concern timing, risk and the power to enforce accountability.
- (29) Truth Has an Operating Cost
Every company needs a story about what is happening and where it is going. The danger begins when that story becomes cheaper to defend than to test.
- (28) The Narrative Premium Eventually Becomes a Reality Discount
Every company needs a story about how it creates value. Trouble begins when preserving that story becomes more important than testing whether it remains true.
- (27) The AI Maturity Gap Is an Operating-Model Gap
AI leaders turn each deployment into reusable institutional capability. Laggards keep funding isolated pilots that must rebuild the path to production.
- (26) Vendor Financing Pulls Demand Forward
AI suppliers are investing in customers, guaranteeing leases and buying unused capacity. Those arrangements can build a market faster, but the market becomes durable only when outside customers replace the supplier's support.
- (25) Local AI Will Make the Cloud More Valuable
Perplexity’s Portable Computer moves routine agent work onto owned hardware and sends only the difficult tail to frontier models. That makes cloud inference more selective, not less valuable.
- (24) Mixture-of-Experts Models Fit the Shape of the Enterprise
Sparse models promise broad capacity at bounded compute, but corporate value depends on memory, explicit controls, and measured customisation.
- (23) How Datacenter Finance Turns Leases Into Bonds
A datacenter is financed several times before it becomes a bond. The sequence works because each stage removes a different risk.
- (22) Multi-Agent AI Systems Are an Architectural Bet Most Companies Are Not Ready to Make
MIT research finds 95% of enterprise AI pilots show no measurable P&L impact, the Bank of England has named AI a systemic financial-stability risk, and reliability math shows a 20-step agent workflow succeeds only 36% of the time. The evidence does not suggest caution; it demands it.
- (21) How to Brief an Agent: The Lost Art of Structured Delegation in Knowledge Work
Frontier models now perform within a narrow band of each other on reasoning benchmarks. The remaining quality gap in agentic AI is not in the models; it is in the human capacity to brief them properly.
- (20) Why AI Adoption Depends on Management
MIT, McKinsey, and BCG each find that only a small minority of firms generate material AI returns. Their advantage is organisational.
- (19) Beyond Retrieval: Recombination as Productivity Gain
Finding an old note saves time. Bringing several notes into a useful new relationship creates something that was not in the archive before.
- (18) Karpathy's LLM Wiki Solves the Maintenance Problem
The useful change in Karpathy’s LLM wiki is not retrieval. The agent maintains the synthesis so the archive can compound between sessions.
- (17) What Self-Hosted Agents Need to Become an Operating System
OpenClaw’s architecture supports the operating-system metaphor. Its isolation model, economics, and evidence base are not there yet.
- (16) AI Is Closing the Entry-Level Hiring Tap
The early labour-market effect is visible in fewer hires rather than mass layoffs. That saves cost now and weakens the future talent pipeline.
- (15) The One-Person Agent Team in Finance
For document-heavy and monitoring-intensive work, one experienced professional can supervise a small agent team—provided the verification layer is real.
- (14) Why Small Banks Struggle to Access SRT
More than 100 banks have issued SRT, protecting nearly €800 billion of loans. The market remains concentrated among large, repeat issuers.
- (13) Why ABF Is Taking More of Private Credit's Growth
Direct lending still holds the larger stock of assets. ABF is taking a disproportionate share of new growth through insurance capital and repeat origination.
- (12) Strategy Fails When the Operating System Contradicts It
A strategy loses force when the organisation’s incentives and decisions teach people that leadership’s words are not a reliable guide to action.
- (11) The Warehouse Trap for Fintech Lenders
A £200 million warehouse announcement says little about control. The decisive terms—eligibility, covenants, amortisation, and exit—sit deep in the contract.
- (10) Where the Agent Layer Captures Value
Cheaper models move the bottleneck upward: value sits in the layer that turns capability into supervised work across tools and systems.
- (9) In Distress, Collateral Data Becomes a Control Asset
Asset-based lenders may own the security interest while the borrower still controls the data that tells them what the collateral is.
- (8) A Practical Case for Shared Collateral Registries
A shared ledger can show that an identified asset was pledged twice. It cannot prove that the asset exists or that its data is accurate.
- (7) Why Boards Underprice Succession Risk
A succession plan is not the same as readiness. Markets discover the difference when a leadership break forces the issue.
- (6) The Governance Cost of Diversification
Diversification can create value, but every added business raises the burden on capital allocation, accountability, and management attention.
- (5) Return-to-Office Mandates Are About More Than Productivity
Blanket office mandates make more sense as governance decisions about control, bargaining power, and property exposure than as productivity policy.
- (4) Consumer Duty Is Becoming a Funding Discipline
Fair-value assessments, outcomes monitoring, and board attestations turn conduct governance into evidence that capital providers can underwrite.
- (3) Why Boards Overweight Strategy and Underweight Execution Risk
Strategy arrives as a paper and a decision. Execution risk builds between meetings, inside controls, systems, vendors, and unresolved dependencies.
- (2) What Venture Equity Actually Costs
Equity creates no monthly payment, which makes it easy to misprice. Its cost arrives later through ownership, control, and return expectations.
- (1) Europe's AI Sovereignty Depends on Its Capital Markets
Europe has the savings, research base, and ambition to build frontier AI. Its capital markets still struggle to fund the scale.