By the ThesisLoop team · Updated 2 Jul 2026
LLM Investment Research Workflow: From Prompts to Process
Move from ad-hoc prompting to a repeatable LLM research pipeline: grounding, retrieval versus long context, and citations at generation time.
Informational research only. ThesisLoop is not investment advice, a stock recommendation, or a guarantee of returns.
Who this page is for
Technical investors, analysts, and builders systematizing LLM-assisted research
Example assets to start with
Why this matters now
Analysts and builders are outgrowing one-off prompts and searching for the architecture decisions that separate reliable pipelines from confident noise.
ThesisLoop research prompt
Design a repeatable LLM research pipeline for [company]: ground on primary documents, choose retrieval or long context per document size, enforce citations at generation time, and add contradiction and staleness checks before any finding is saved.
Start with this promptEvidence checks
Ground generation on primary documents rather than web search, and record which documents were in scope for each output.
Choose retrieval versus long context deliberately for 200-page filings, and test whether key sections survive the choice.
Enforce citations at generation time — a claim without a document and page reference should fail the pipeline, not ship.
Run contradiction and staleness checks across documents and periods, and evaluate outputs against the sources, not against other model output.
Research questions
What grounding corpus is sufficient before the pipeline runs — filings, transcripts, decks, or all three?
Where does retrieval lose material context on long filings, and where does long context lose precision?
How are contradictions between documents surfaced instead of silently averaged away?
Is building this pipeline the best use of research time, or is adopting the workflow the better trade?
Public report examples
Use these published reports as examples of source-backed research structure: claims, evidence, risks, and follow-up questions. They are educational examples, not investment advice or recommendations.
Keywords this page covers
The goal is not a keyword list. The goal is to turn a search query into a specific, source-backed research workflow.
Related research topics
Move from a broad theme into adjacent company-level diligence.