🔒 Fidelity check is locked

This audio↔story breakdown is insider-only. If you know the key, add it to the URL — if not, ask around. 😉

← back to the story

🎧 Audio ↔ 📄 Story · Fidelity Check

How close is the page to Venky's voice?

Comparing the 16:45 recording (auto-transcribed, ~2,688 words) against the published The Space Between (~1,389 words).  ·  ← back to the story

0/100faithfulness
Faithful retelling — the whole recording, tightened.

The page tracks Venky's recording beat for beat, start to finish — the cancelled March workshop, Sri's vulnerability, the two brands at different speeds, the evolution-vs-revolution question, the airport call. The dates and quotes hold. It's compressed to about half the spoken length and re-worded for reading — not invented.

Where the score comes from

Story beats preservedDid every moment Venky told make it onto the page?88%
Facts & quotes accuracyNumbers, names, dates, and direct quotes85%
Tone & voice matchDoes it still sound like Venky telling it?90%
Verbatim wordingWord-for-word overlap — low by design for a written piece17%
Length retainedStory is 52% the length of the spoken version52%

▮ green = meaning preserved · ▮ amber = minor drift · ▮ grey = expected difference (rewrite, not loss)

✓ Kept faithfully

  • The cancelled March workshop — March 2025, "quite a gloomy moment"
  • Sri's vulnerability — "I haven't earned the right to coach others on AI yet… do it inside one team first" (near-verbatim)
  • Devika & Voice IQ — belief travelling both ways; James + AI proposal generation from call transcripts
  • "The goal feels aggressive… but I trust you" — Devika's candid line
  • The 2024 seed idea — train AI on the last few hundred sold proposals (Neelesh / "The Spark")
  • Anish's 50% framing — halve proposal time, give it back for customers and selling
  • Europe Express parallel track — Casey & Carol, "DIY → do-it-for-me", and the two brands at different speeds
  • Cookie-cutter / blue-sky principle & the canvases defending ET's own discovery
  • Evolution vs. revolution — Sri's question, and the "script history" Slack channel
  • July 16 engineering foundation — daily pairing with Sreenath; AI-first discovery heresy; "cut the noise"
  • Alex's challenge — "five days and all we get is a hypothesis"
  • Rajesh's private "I'm a bit lost" message, the airport call, "I know what I know", eight points of alignment

~ Changed / dropped

  • "Act one — early spark and initial confusion" framing dropped; page opens on "the boring middle"
  • March 3, 2025 exact date generalised to "March 2025"
  • "eGenAI proposal gen" PPD title not named — page says only "same blueprint"
  • HubSpot / "top disc" / Ustara tool names → genericised to "their own tooling"
  • James & Matt "calling it a blessing" (Voice IQ go-live) — trimmed
  • Anish's ordering (guest-preference automations first, email/SMS second, proposals third) — collapsed
  • "design sprint" / two-day discovery aside — compressed into the AI-first heresy
  • Artika / WordPress-partner visit — page says only "to visit a partner"
  • Compression: page is ~52% of the spoken length — recurring "right?" asides and repeats are smoothed out
Read it like this: the big number is faithfulness of meaning, not transcription accuracy. A score in the high 80s means "trust this page as Venky's story." The 17% verbatim and 52% length bars are expected — this is a long, looping spoken recording tightened into a readable voice, not a transcript.

Method. Audio transcribed locally with whisper.cpp (base.en). Verbatim (17%) and length (52%) are measured directly (difflib sequence ratio & word counts). Beats, facts, and tone are a beat-by-beat human/AI reading of transcript vs. page. Overall faithfulness is a weighted blend (beats 40% · facts 35% · tone 25%); verbatim is reported separately, not folded in.