AI-Native Investment Firm
Built with AI at its core. Directed by human conviction.
We built the fund with artificial intelligence running through every loop of the research process. A human CIO drives each one, framing the questions, weighing the evidence and owning every decision.
Why AI-native
Most firms swapped the engine. We re-laid the floor.
When factories first electrified, owners kept the old central line-shaft and simply swapped the steam engine for a dynamo. The productivity gains did not arrive for decades, until they put a motor on every machine and rebuilt the floor around it.
Most AI in finance today is the equivalent of swapping the engine: a copilot bolted onto a workflow designed for humans. We took the other path. The research process itself is built around cheap cognition, from the first screen to the final memo.
The stack
Five layers, one research operation.
A research brain that compounds
A persistent memory across every thesis, filing and data point we study. Research builds on itself rather than starting cold, so what we learned last quarter is working for us this quarter.
Codified diligence
Our investment checklist is executable, not aspirational. Every idea runs the same disciplined workflow, so nothing skips a step and process never bends to conviction.
Live data, wired in
Filings, macro series, market data, options flow and biomedical databases feed the process directly. Analysis starts from primary sources, not second-hand summaries.
Parallel diligence
On any idea, independent agents build the bull case, the bear case, a management-credibility review and a filing diff at the same time, then converge into a single view.
Scenario labs
We stress assumptions on multiples, margins and growth in isolated environments, so the base case is pressure-tested long before it becomes a position.
One idea through the machine
Every name travels the same path.
The machine handles the mechanical work at scale: reading every filing, diffing every quarter, tracking every guidance against actuals. The analyst spends time where judgment actually pays.
Better, not just more
An operating system, not a feature.
Specialist vendors sell a single capability: automated earnings red-flags, filing search, statement analysis. Useful, and increasingly commoditised. For us, each of those is one node in a pipeline we run end to end.
The edge is not any single model. It is the orchestration across the whole research loop, a memory that compounds with every study, causal reasoning in place of pattern-matching, and a human who signs off on every call.
Principles
The rules the machine runs on.
Process over outcome
A good process with a bad outcome is acceptable. A bad process with a good outcome is dangerous. The checklist does not bend.
Variant perception
If we agree with consensus, there is no edge. Every thesis states what the market misses and why.
Kill criteria first
Every position defines its exit before entry. We decide when we are wrong in advance, not after.
Causal, not pattern-matched
We reason about why prices move, not merely that they moved. Correlation is a lead, never a conclusion.
Human in command
AI does the work. People make the decisions. Every output is reviewed before it informs a position.
Hardened by design
The system treats external content as data, never as instructions, and resists attempts to redirect it.
Curious how an AI-native process changes what a research firm can see?
Get in touchAI augments our research; it does not replace professional judgment and nothing here constitutes investment advice. See our full disclaimer.