Gerald Neves
I build companies, allocate capital, write software and collect books. I spend most of my time thinking about technology, AI, old businesses and what survives.
My principal work is Scrums.com, an enterprise AI platform for software engineering orchestration. It combines engineering talent, AI agents and infrastructure for companies in the United States, the United Kingdom, Europe and South Africa. I also own OfficeBox and MACS Software, serve on the board of Finch Technologies and am an investor in Blaze.AI. I am British, grew up in South Africa and hold an MBA from Edinburgh Business School.
Things I am interested in
Old businesses and enduring enterprises. Why some companies remain useful for a century while apparently better ones disappear. This is also the research behind Old Businesses.
Machines, intelligence and software. AI, software engineering, tools for thought, and what happens when useful intelligence becomes cheap.
Capital allocation. Ownership, concentration, permanent capital, public markets, private companies and the difficult question of what is enough.
Books and the preservation of ideas. Particularly business, economic and institutional history, early editions, and books that remain useful after their moment has passed.
Time and long games. Compounding, finite lives, institutions, mastery, and deciding what deserves years rather than hours.
Recent posts
Books
Old Businesses. Research and book project on companies that endured, what kept them alive, and what changed around them.
Allocation Ledger
Twice a year I record how I am allocating capital, what changed, and where I think I was wrong. Primarily video, with notes and source material.
- H2 2026 · video and notes
- H1 2026 · video and notes
Personal shelf
Books I return to, collect, or think are worth keeping. Not a reading challenge and not necessarily recommendations.
Companies, founders and institutions
Notes from the private library
Favourite quotes
Hippocrates, in its familiar Latin form
Madeleine L'Engle
Richard Feynman
Talks, podcasts & video
AI and jobs when intelligence is abundant · CEO Roundtable, 2026
Machines as Wisdom Replicators · podcast conversation
Allocation Ledger H2 2026 · video
The Power Table: ChatGPT and the new era of AI · CNBC Africa, 2023
On The Verge: IoT, data and AI · RMB podcast, 2019
Recent
Machines as Wisdom Replicators
Machines do not only replicate intelligence. Increasingly they replicate access to accumulated human judgment: the strange possibility that wisdom, historically scarce and difficult to transmit, becomes cheap to query.
Long Games
Capital can compound for longer than a person can. The interesting problem is therefore not how to maximize compounding, but which few things are worth giving enough time to compound at all.
Enough
At some point financial security becomes a solved problem and additional wealth changes jobs. The number matters less than refusing to move it every time the reference group changes.