Analysis that lives exclusively inside a chat thread does not travel well. If you want to pitch an investment committee, share findings with a mentor, or simply leave a reliable breadcrumb trail for your future self, you need structure. You need mind maps that hold a complex thesis together, comparison views that make trade-offs immediately visible, and scenario outlines that force you to define what must go right—or wrong—over a five- to ten-year holding period.
Your job in this phase is to act as the editor: ensuring every node, bubble, and bullet point maps back to hard financial tables and your own conviction, rather than slipping into fluent, AI-generated generalities.
From Analysis to Artifacts
Once you have completed the rigorous durability and valuation checks outlined in Part 3, you can prompt Gemini to convert the underlying work into three distinct presentation layers:
- The Long-Term Thesis Map (One per company)
- The Comparative View (Across the entire theme or sector)
- Scenario Slides (Base, Bull, and Bear cases over 5–10 years)
Treat these artifacts as communication tools first, and memory tools second. They serve as a structural safeguard, reducing the chance that a compelling qualitative narrative drifts away from the quantitative numbers that initially justified it.
Key takeaway: Visuals and slide outlines are not decoration—they are a coherence test. If a branch in your mind map cannot cite a specific metric or filing claim, delete the node.
1. Long-Term Thesis Maps (Per Company)
A mind map forces a linear narrative into a holistic, interconnected view. Ask Gemini to generate a hierarchical outline that you can easily copy and paste into your preferred mapping tool (like Whimsical, XMind, Miro, or simply a nested Markdown file).
Force the AI to stick to these fixed branches:
- Business Model: How the company makes money, customer mix, and geographic footprint.
- Moat: The specific type of advantage, cited evidence, and durability concerns.
- Financial Track Record: The 10-year shape of ROE, margins, FCF, and dividends.
- Capital Allocation: Reinvestment rates, M&A history, buybacks, and debt trajectory.
- Valuation: Current multiples versus history and peers (including your explicit assumptions).
- Key Long-Term Risks: Structural, operational, financial, and governance threats.
Prompt 1: The Thesis Mind Map Outline “Create a mind-map outline for [TICKER] with the following main branches: Business Model, Moat, Financial Track Record, Capital Allocation, Valuation, and Key Long-Term Risks. Under each branch, provide 3–6 concise nodes. Every single node must explicitly reference either an uploaded filing (with the year), a data point from the metrics-summary table, or my original thesis note. Mark any nodes that are unproven hypotheses with [H]. Do not add generic industry filler.”
Optional Discipline: Once you paste the outline into your mapping tool, color-code the nodes. Use green for evidenced strengths, amber for mixed historical evidence, and red for structural risks that could invalidate the entire investment.
2. The Comparative View (Across Companies)
Long-term stock selection is a game of opportunity cost; everything is relative. You need a visual framework that places your candidates side-by-side across three dimensions:
| Dimension | Typical Inputs |
|---|---|
| Quality | Moat evidence + ROE consistency over a decade. |
| Durability | Free Cash Flow trajectory + leverage and debt behavior through cycles. |
| Valuation | Current multiples relative to each name’s 10-year history and peers. |
Prompt 2: The Comparative View “Using only the materials uploaded to this notebook, score each candidate company from 1 to 5 on three dimensions: Quality (moat evidence + ROE consistency), Durability (FCF and leverage behavior), and Valuation attractiveness relative to its own 10-year median (where 5 = cheapest vs. history). Provide a one-sentence justification for every score with citations. Finally, draft a single comparison slide outline featuring three columns (Quality / Durability / Valuation) and a closing summary highlighting the most critical trade-offs a long-term holder must weigh between these companies.”
Warning: Resist the urge to let the AI average the scores into a fake “overall winner.” The entire point of the comparative view is to make disagreements and trade-offs visible (e.g., choosing between an expensive, ultra-high-quality compounder versus a cheap, highly durable cyclical stock).
3. Scenario Slides: Base, Bull, and Bear (5–10 Years)
Scenario planning prevents single-path optimism. A stock held for a decade will inevitably face macro shocks, management changes, and strategic pivots. Have Gemini draft scenario slide outlines so you know exactly what to watch for over the years.
- Base Case: Central assumptions for volume, margins, and historical capital allocation discipline.
- Bull Case: What must go right (e.g., successful market adjacency, strengthening moat) and which major risks fail to materialize.
- Bear Case: Structural breaks (regulation, disruption, customer loss) and leverage/dividend stress that would force you to abandon the thesis.
Prompt 3: Scenario Deck Outline “Draft a presentation slide outline detailing 5–10 year scenarios for [TICKER]. Create three slides: Base, Bull, and Bear. Each slide must include: (1) 4–6 key business drivers, (2) 3 specific financial signposts to monitor annually (e.g., ROE, FCF/share, net leverage), (3) 3 risk factors, and (4) the specific triggers that would confirm or falsify this scenario. Tie all drivers to historical evidence already in the notebook. Avoid hype adjectives. End with a single ‘Annual Monitoring Checklist’ slide common to all scenarios.”
4. Review for Coherence (The Human Audit)
Large Language Models want to please you. Gemini will occasionally promote management’s marketing language into “moat facts,” smooth over bad financial years to make a narrative look cleaner, or invent neat symmetry across peers where none exists.
Before finalizing your presentation, run a strict human review against this checklist:
- Numbers: Do the map nodes and slide bullets match the 10-year tables exactly?
- Citations: Can you open the cited filing year and easily find the claim?
- Balance: Do the risk branches receive the same level of analytical rigor as the strength branches?
- Conviction: Does the final narrative match your initial thesis note, or did the model subtly rewrite your beliefs?
If you find discrepancies, do not just fix the cosmetics on your slide. Run a coherence audit prompt to discipline the AI.
Prompt 4: The Coherence Audit “Compare the mind-map outline and scenario slides you generated for [TICKER] against the 10-year metrics summary tables and filings. List every statement in your output that is unsupported, contradicted by the numbers, or overgeneralized. Propose specific edits to fix these errors. Do not defend prior wording; just correct the inaccuracies based on the hard data.”
The Practical Build Sequence
To execute this phase efficiently, follow these steps:
- Generate a thesis mind-map outline per company and import it into your visual tool.
- Build the comparative Quality / Durability / Valuation slide across the full peer set.
- Draft Base / Bull / Bear slide outlines focusing on drivers, signposts, and falsifiers.
- Run a coherence audit prompt to catch AI hallucinations or narrative smoothing.
- Edit the maps and slides manually, re-prompting only where the gaps are obvious.
- Store these final text outlines back into the Gemini Notebook as reference documents, so future updates start from a known baseline.
If an AI-generated visual cannot survive contact with the underlying numbers, it is a story, not an analysis. Long-term investors need the latter.