Key Takeaways
- Paul Enright separates equity research by knowledge half-life, arguing that outsourcing initial company work to large language models robs analysts of permanent mental models.
- Offloading initiation memos or financial modeling to tools like Claude creates cognitive debt, leaving an investor unable to understand the reflexivity between business drivers and stock performance.
- Routine maintenance tasks, like updating models after quarterly earnings calls or tracking consensus shifts, should be automated because that data loses its value within days.
- Letting algorithms choose key performance indicators blinds an investor to how a company actually generates cash.
- The distinction between manual initiation and automated maintenance forms Enright's Half-Life Research Workflow.
The Enright's Half-Life Research Workflow
Former Viking Global partner Paul Enright watched junior analysts start leaning on large language models to write company initiation memos. The speed looked impressive, but the retention was empty. An analyst who generates a three-page summary of an SEC filing in Claude cannot defend the thesis under fire. They borrow the conclusion without doing the cognitive lifting.
Enright calls this cognitive debt. When you skip reading the primary documents, you miss the footnotes, the subtle changes in management tone, and the operational bottlenecks. As Enright warns:
“Avoiding cognitive debt is super important right now. If you go and say, 'Oh, I did a screen on AI, I have Claude delivering this summary. Claude wrote the initiation memo for me on XYZ company.' I guarantee you, you won't embed it in your memory and remember it the same way if you did it yourself.” (play from 30:44)
To preserve real conviction while capturing machine speed, Enright draws a strict line between initiation and maintenance:
- Step 1: Launch Coverage Manually (Build Long Half-Life Crystallized Knowledge): Do all initiation work personally without AI shortcuts. Read every primary document, read background books, walk factory floors, test products, build financial models manually, and select the specific key performance indicators (KPIs) to understand the reflexivity between inputs and outputs.
- Step 2: Automate Maintenance and Ephemeral Tasks (Short Half-Life Delegation): Automate high-velocity, short half-life tasks using AI and data pipelines. Use automation to track shifting consensus earnings estimates, generate macro rate dashboards, and update model financial statements within seconds of earnings releases.
Building the spreadsheet line by line is where an investor spots the friction. As Enright explains: “Don't let AI decide what the inputs are because you won't understand the inputs. You won't understand the reflexivity between the inputs and the outputs if you haven't done it yourself.” (play from 31:10) Once that foundation exists, spending hours hand-entering earnings press releases is an error. “The stuff that you want to automate and that you want to give to AI is the stuff that has a pretty short half-life where it's a complete waste of your time,” (play from 30:00) Enright says. “The process of launching coverage, do it yourself. The process of maintaining coverage, get the help.” (play from 31:29)
When This Works (and When It Doesn't)
This framework applies to research analysts and knowledge workers who need to develop deep, crystallized domain expertise while maximizing productivity and avoiding cognitive debt. It shines in concentrated portfolios where an investor owns twenty companies and needs to know them cold. In high-conviction investing, underwriting error is fatal. Knowing which two KPIs drive operating leverage lets you take a stand when the street panics.
This framework breaks down for quantitative strategies and short-horizon statistical arbitrage. If your edge relies on trading thousands of earnings revisions across five hundred tickers, building bespoke models by hand is a liability. It also fails in pure early-stage venture screening, where high deal volume requires shallow pattern matching across hundreds of pitch decks before deciding where to spend real diligence hours.
Why It Matters
As multi-manager platforms compress investment horizons, many teams mistake fast retrieval for deep understanding. Analysts who rely on automated digests look competent in calm markets, but they break during unexpected earnings misses because they never internalized the operating plumbing. Capital allocators who evaluate manager edge will increasingly differentiate between shops that possess genuine domain mastery and those carrying hidden cognitive debt.