The investment process is the product
Asset managers sell a repeatable process. As research and operations adopt AI, that process has to become explicit enough for machines to follow and for clients to verify.
By Penomic Research · Published · 7 minute read
Download the PDFClients buy the process
Institutional clients and consultants select managers on the strength and consistency of their investment process. Due diligence questionnaires, investment committee minutes and portfolio construction rules all try to describe it. In reality much of the process lives in how portfolio managers and analysts interpret those rules day to day.
When AI enters research and operations, that gap matters. A model that summarises earnings calls or screens securities will apply a generic view of quality and value unless it is given the firm's own.
The knowledge behind a house style
A manager's house style is built from many small, explicit and implicit choices.
- Research conventions: which adjustments analysts make to reported figures and why.
- Quality and valuation criteria and how they trade off against each other.
- Portfolio construction rules, risk limits and the circumstances that justify exceptions.
- Sell discipline and the signals that trigger a review.
- Lessons from past mistakes that changed the process.
Consistency is now a client question
Allocators increasingly ask how managers use AI and how they keep it within the stated process. A manager that can show which definitions, rules and evidence an AI-assisted recommendation relied on has a stronger answer than one that relies on analyst review alone.
The same structure helps internally. When a key portfolio manager leaves, the process they ran remains available in a governed form instead of in their notebooks.
Where to start
Choose one strategy and one recurring task, such as the initial screen or the quarterly review of holdings. Write down the conventions and thresholds the team actually uses, validate them against recent decisions, and run AI-assisted analysis within those rules with provenance on every output. The outcome is both better research support and a clearer story for clients.