Straight Answers
An LLM assessed 668 question → answer exchanges, the host's own answers included. Speaker names were hidden during assessment. Each exchange was marked answered, partly, redirected or not addressed, with a quote as its source. Every exchange below is timestamped and plays. How it works →
These readings are observational: the transcript shows whether the question was answered directly, whatever the reason. It is an AI reading of a public exchange, never a claim about intent, and the tape is one tap away on every row. full disclaimer →
Every one of the 330 people on the record is accounted for: 8 carry a full rate (8 or more assessed questions on raw tape), 154 more have their questions counted (21 of them with at least one not answered directly; too few raw questions for a fair percentage, edited feed included), and the remaining 168 never took a direct question in any tape we hold: compilations, panels and cameo appearances leave nothing to assess.
Alessio Fanelli: every exchange, playable
The question asked, then the AI assessment note from the answer. Tap to play the moment. Not addressed = the question was not engaged; redirected = acknowledged, then steered elsewhere.
“But then I think from there, the actual exploration and making that great, that is a hard design task still. So how do we lower the floor for everyone coming in, but also raise the ceiling, make us the designers can do even more and produce even greater work?”
answered: “So I'm curious, like, if there's something that you think about”
“Basically, that's really the key question here. So optimizing the embedding model, um, even changing the way you like chunk things, these all shift the embeddings.”
answered: “So the retrieval is interesting. I got a bunch of startup pitches that are like”
“I mean, it underlines it if you reach a 150 tokens or a 150,000 tokens or something. How do you teach this to the user?”
answered: “we're in the church of context engineering at the Chrome office.”
All speakers
All 8 people with 8 or more assessed questions on raw tape. Click a name to see their exchanges. highest rate none identified small samples
| Person | Not answered directly (%) | Redirected + not addressed | Questions | Argument clarity /5 |
|---|---|---|---|---|
| Alessio Fanelli Partner & CTO, Decibel | 30% | 3 | 10 | 3.7 |
| Joon Sung Park Co-founder & CEO, Simile | 20% | 2 | 10 | 4.3 |
| Emily Glassberg Sands Head of Information, Stripe | 13% | 1 | 8 | 4.3 |
| Dharmesh Shah Co-Founder and CTO, HubSpot | 10% | 1 | 10 | 4.1 |
| Shawn Wang Cofounder & CEO, AI Engineer | 7% | 1 | 15 | 3.9 |
| Michelle Pokrass Post-Training Research Lead, OpenAI | 0% | 0 | 9 | 4.4 |
| Erik Schluntz Member of Technical Staff, Anthropic | 0% | 0 | 9 | 4.4 |
| Thomas Scialom Senior Staff Research Scientist, Meta AI | 0% | 0 | 12 | 4.2 |
Percentages come from raw unedited recordings only: produced podcast audio has tangents and stumbles cut in the edit, which moves the speaker up by about 12 percentile points. The small-sample counts include the produced feed. This is a reading of a transcript. It is not an accusation of dishonesty, and there are many good reasons not to answer a question directly (confidential numbers, unreleased products, someone else's news to break).