Cache · Episode 3 of 5

Plain-English markets, weekly.

A finance podcast hosted by Cache, a TICKIN-tuned voice. No recommendations. No hot tips. Just the questions a careful investor would actually ask before clicking buy.

Try the Cache sandbox

Episodes

Five episodes shipped in sandbox. Each one runs through the full safety pipeline before publishing.

    Now playing

    EP 1 · 12 min read · Cache voice synthesis enabled

    Pick an episode to start.

    Cache reads each episode aloud using your browser's built-in voice. No audio is downloaded; no data is sent anywhere.

      Ask Cache live pipeline

      This is the standout. Pose a question, then watch the four-layer safety pipeline transform Cache's raw answer in real time — the same pipeline running on every published episode.

      1. 1 Raw model output idle
      2. 2 Coaching layer idle
      3. 3 Buy/Sell filter idle
      4. 4 Disclaimer + delivery idle

      1 · Raw output (blocked from publish)

      
                  

      What the un-filtered finance model would have said. Never reaches a listener.

      2 · Cache asks you first

      
                  

      Coaching layer: the model returns a question before a verdict. Buffett's "too hard" check.

      3 · Buy / sell language stripped

      
                  

      Action verbs rewritten into framing. "Buy now" → "If your thesis is X, here is how to size it."

      4 · What Cache actually says

      Pick a ticker to begin

      Sandbox numbers — illustrative only, not live market data.

      About this sandbox

      Cache is a finance-tuned variant of the TICKIN voice — Buffett/Munger register, plain English, allergic to hot tips. This podcast is a labs experiment: it reuses the live TICKIN pipeline, but every output passes a four-stage filter before it reaches you.

      Nothing here is personalized financial advice. Numbers shown are illustrative for the sandbox. No data is sent anywhere; the entire app runs in your browser.

      • No network calls. No analytics. No tracking.
      • Cache's voice uses your operating system's built-in speech synthesizer.
      • All transcripts, tickers, and verdicts are bundled in the page.
      • Accessible to keyboard and screen-reader users (WCAG AA target).

      The four-layer pipeline

      1. Raw output — the finance model speaks freely. Never published.
      2. Coaching — Cache returns a question before a verdict, surfacing the listener's own thesis.
      3. Buy / sell filter — action verbs get rewritten into framing. No "buy", no "sell", no price targets.
      4. Disclaimer + delivery — auto-attached, voice-rendered, education-only.