Grade My Investments Technology Observatory
GMI · TECHNOLOGY OBSERVATORY // ALL SYSTEMS NOMINAL
ENGINEERED BY LEOPARD DATA

The AI-Augmented Investor Workflow

A prescriptive recipe: how to combine a general AI assistant with GMI's graded hard data into one evaluation loop — yes, we're telling you to use other AI tools, because the loop comes back stronger

THE LOOP
rendering diagram…
flowchart LR
    OWN[What you own:<br/>funds, ETFs, stocks] --> AI1[General AI:<br/>decompose into<br/>a symbol list]
    AI1 --> GMI[GMI: grade every symbol<br/>on audited fundamentals<br/>published algorithm]
    GMI --> TIP[Triage by grade:<br/>D's and F's, surprising A's,<br/>grade-vs-ML disagreements]
    TIP --> ASK[Ask Claude inside GMI:<br/>interpret YOUR report's data]
    TIP --> AI2[General AI:<br/>articles, filings coverage,<br/>competitive narrative]
    ASK --> DEC[Human decision]
    AI2 --> DEC
    DEC --> SHARE[Share the report link<br/>free, no account needed]
    DEC -. rerun after earnings,<br/>before adding money .-> GMI
AI at both ends, audited hard data in the middle. Each tool does what it's actually good at: LLMs decompose and interpret; GMI grades on numbers with provenance; you decide.

The Recipe, Step by Step

  1. Decompose what you own with a general AI. Ask Claude (or your assistant of choice): "Give me a comma-delimited list of the top holdings of [your fund/ETF]." Funds are where most people's money actually lives, and this one prompt turns an opaque ticker into a concrete list of companies. (For supported ETFs, GMI's holdings import does this natively.)
  2. Paste the list into GMI and run a report. Every symbol gets graded on audited fundamentals — the published algorithm over numbers with provenance. This is the step a general AI cannot do reliably on its own: LLMs paraphrase numbers; GMI's are computed from filings.
  3. Let the grades tip you off. Sort by grade. Your shortlist is the D's and F's (what decayed?), the surprising A's (what's quietly excellent?), and any disagreement between the deterministic grade and the ML health score. Minutes of triage replaces hours of skimming — and the known blind spots tell you which grades to double-check rather than trust.
  4. Interrogate the shortlist — with AI on both surfaces. Inside GMI, use Ask Claude against the report itself: "why is this health score falling?" "compare free-cash-flow trends across my chip names." It answers from your actual data. Then take the flagged names back to a general AI for the qualitative layer — recent articles, filings coverage, competitive narrative — the context that numbers can't carry.
  5. Decide like a human, document like a pro. The tools produce a shortlist and a dossier; the decision stays yours. Grades inform, never dictate.
  6. Share the conclusion, rerun on cadence. Share the report link with whoever you talk stocks with (it's free, they need no account). Then rerun the loop when it matters: after earnings seasons for holdings, before adding money, and whenever a grade-drop notification or gut instinct says look again.

Why We Send You to Other AI Tools Mid-Workflow

Because it's honest, and because the loop returns. General AI assistants are genuinely the best tool alive for decomposition and open-web synthesis — and genuinely unreliable as a source of financial figures, where confident hallucination is the failure mode. GMI is the inverse: authoritative on the numbers, and it embeds Claude grounded in those numbers for the interpretive step. The workflow that uses each where it's strong beats either alone — and this is exactly how the founder runs his own real portfolio, funds decomposed by AI, graded by GMI, investigated by both.

One rule keeps the loop safe: numbers flow from GMI to the AI, never the other direction. Ask an LLM to interpret graded data and it excels; ask it to supply the data and you're building on sand.