StarZero All shows Try the podcast clipper →

Nov 20, 2014 · 29m · mad

Dennis Mortensen and Marcos Jimenez, x.ai // Your AI-powered Virtual Assistant

Marcos Jimenez · 13m spoken Dennis Mortensen · 9m spoken Matt Turck · 47s spoken Sol Weinreich · 36s spoken L.A. Noma · 18s spoken Alexander Koch · 12s spoken Laura Teller · 11s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

At a DataDrivenNYC presentation, x.ai founder Dennis Mortensen and Chief Data Scientist Marcos Jimenez introduce 'Amy,' an AI virtual assistant that automates meeting scheduling via plain-English email interactions. They outline the company's product vision alongside the natural language processing, intent classification, and user preference algorithms that make autonomous scheduling possible.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 3.3% of the talking time here. How this is scored →

Matt as informed peer 0.6 Guest teaching 3.0 Guest disagreement 0.6 Matt pushing back 0.4
05100:0010:0020:003:04–5:43 · Matt as informed peer 0/10 x.ai Vision and AI Automation Approach Dennis Mortensen delivers a presentation monologue describing x.ai's vision and strict refusal to rely on human backup plans. The host is absent during this presentation segment, requiring host-side scores of zero.5:43–11:54 · Matt as informed peer 0/10 x.ai Architecture & Guided Conversation Model Marcos Jimenez presents the technical architecture behind Amy, clarifying that humans serve as AI trainers doing quality assurance rather than direct virtual assistants. The host does not participate during this presentation block.11:54–16:07 · Matt as informed peer 0/10 Natural Language Processing & Email Intent Categorization Marcos outlines natural language processing models, 25 intent categories, and the deployment of Support Vector Machines (SVMs) with plat scaling. The host is not involved, resulting in zero host scores.16:07–19:04 · Matt as informed peer 0/10 Information Extraction, User Preference Analysis & Conclusion Marcos concludes the deck by explaining user calendar preference analysis and probability distributions for scheduling. Host interaction remains absent throughout the technical presentation.19:04–29:20 · Matt as informed peer 3/10 Audience Q&A with Dennis Mortensen and Marcos Jimenez Host Matt Turck opens Q&A citing his experience as a beta user and presses Marcos on the boundaries of automation without major data science breakthroughs. The dialogue throughout the audience Q&A remains highly collaborative and informative.3:04–5:43 · Guest teaching 2/10 x.ai Vision and AI Automation Approach Dennis Mortensen delivers a presentation monologue describing x.ai's vision and strict refusal to rely on human backup plans. The host is absent during this presentation segment, requiring host-side scores of zero.5:43–11:54 · Guest teaching 3/10 x.ai Architecture & Guided Conversation Model Marcos Jimenez presents the technical architecture behind Amy, clarifying that humans serve as AI trainers doing quality assurance rather than direct virtual assistants. The host does not participate during this presentation block.11:54–16:07 · Guest teaching 4/10 Natural Language Processing & Email Intent Categorization Marcos outlines natural language processing models, 25 intent categories, and the deployment of Support Vector Machines (SVMs) with plat scaling. The host is not involved, resulting in zero host scores.16:07–19:04 · Guest teaching 3/10 Information Extraction, User Preference Analysis & Conclusion Marcos concludes the deck by explaining user calendar preference analysis and probability distributions for scheduling. Host interaction remains absent throughout the technical presentation.19:04–29:20 · Guest teaching 3/10 Audience Q&A with Dennis Mortensen and Marcos Jimenez Host Matt Turck opens Q&A citing his experience as a beta user and presses Marcos on the boundaries of automation without major data science breakthroughs. The dialogue throughout the audience Q&A remains highly collaborative and informative.3:04–5:43 · Guest disagreement 2/10 x.ai Vision and AI Automation Approach Dennis Mortensen delivers a presentation monologue describing x.ai's vision and strict refusal to rely on human backup plans. The host is absent during this presentation segment, requiring host-side scores of zero.5:43–11:54 · Guest disagreement 0/10 x.ai Architecture & Guided Conversation Model Marcos Jimenez presents the technical architecture behind Amy, clarifying that humans serve as AI trainers doing quality assurance rather than direct virtual assistants. The host does not participate during this presentation block.11:54–16:07 · Guest disagreement 0/10 Natural Language Processing & Email Intent Categorization Marcos outlines natural language processing models, 25 intent categories, and the deployment of Support Vector Machines (SVMs) with plat scaling. The host is not involved, resulting in zero host scores.16:07–19:04 · Guest disagreement 0/10 Information Extraction, User Preference Analysis & Conclusion Marcos concludes the deck by explaining user calendar preference analysis and probability distributions for scheduling. Host interaction remains absent throughout the technical presentation.19:04–29:20 · Guest disagreement 1/10 Audience Q&A with Dennis Mortensen and Marcos Jimenez Host Matt Turck opens Q&A citing his experience as a beta user and presses Marcos on the boundaries of automation without major data science breakthroughs. The dialogue throughout the audience Q&A remains highly collaborative and informative.3:04–5:43 · Matt pushing back 0/10 x.ai Vision and AI Automation Approach Dennis Mortensen delivers a presentation monologue describing x.ai's vision and strict refusal to rely on human backup plans. The host is absent during this presentation segment, requiring host-side scores of zero.5:43–11:54 · Matt pushing back 0/10 x.ai Architecture & Guided Conversation Model Marcos Jimenez presents the technical architecture behind Amy, clarifying that humans serve as AI trainers doing quality assurance rather than direct virtual assistants. The host does not participate during this presentation block.11:54–16:07 · Matt pushing back 0/10 Natural Language Processing & Email Intent Categorization Marcos outlines natural language processing models, 25 intent categories, and the deployment of Support Vector Machines (SVMs) with plat scaling. The host is not involved, resulting in zero host scores.16:07–19:04 · Matt pushing back 0/10 Information Extraction, User Preference Analysis & Conclusion Marcos concludes the deck by explaining user calendar preference analysis and probability distributions for scheduling. Host interaction remains absent throughout the technical presentation.19:04–29:20 · Matt pushing back 2/10 Audience Q&A with Dennis Mortensen and Marcos Jimenez Host Matt Turck opens Q&A citing his experience as a beta user and presses Marcos on the boundaries of automation without major data science breakthroughs. The dialogue throughout the audience Q&A remains highly collaborative and informative.

speaking balance: gold is Matt, purple is the guest (3 minute bins)

0:00 · Matt 0% · guest 100%0:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%15:00 · Matt 0% · guest 100%15:00 · Matt 0% · guest 100%18:00 · Matt 27.5% · guest 72.5%18:00 · Matt 27.5% · guest 72.5%21:00 · Matt 0.4% · guest 99.6%21:00 · Matt 0.4% · guest 99.6%24:00 · Matt 1.4% · guest 98.6%24:00 · Matt 1.4% · guest 98.6%27:00 · Matt 5.2% · guest 94.8%27:00 · Matt 5.2% · guest 94.8%
Sharpest disagreement ▶ 3:45 Rejection of Human Safety Net Plan B

Dennis forcefully rejects the idea of a human fallback strategy, humorously noting investor anxiety while insisting on an uncompromising data-training approach.

 Hardest push from Matt ▶ 19:03 Host Probes Limits of Pure Automation

Matt Turck gently challenges the 100% target by asking how far x.ai can realistic go in the near term without requiring fundamental advances in data science.

 Biggest teaching moment ▶ 17:20 Calendar Probability Distributions

Marcos educates the audience on how machine learning extracts implicit calendar preferences like meeting density and lunch hours to intelligently place meetings.

 Matt holds his own ▶ 19:03 Host Leverages Beta-User Domain Knowledge

Matt Turck demonstrates informed hands-on familiarity with x.ai's performance as a beta user before framing a precise question on model limitations.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
x.ai Vision and AI Automation Approach 0220 Dennis Mortensen delivers a presentation monologue describing x.ai's vision and strict refusal to rely on human backup plans. The host is absent during this presentation segment, requiring host-side scores of zero.
x.ai Architecture & Guided Conversation Model 0300 Marcos Jimenez presents the technical architecture behind Amy, clarifying that humans serve as AI trainers doing quality assurance rather than direct virtual assistants. The host does not participate during this presentation block.
Natural Language Processing & Email Intent Categorization 0400 Marcos outlines natural language processing models, 25 intent categories, and the deployment of Support Vector Machines (SVMs) with plat scaling. The host is not involved, resulting in zero host scores.
Information Extraction, User Preference Analysis & Conclusion 0300 Marcos concludes the deck by explaining user calendar preference analysis and probability distributions for scheduling. Host interaction remains absent throughout the technical presentation.
Audience Q&A with Dennis Mortensen and Marcos Jimenez 3312 Host Matt Turck opens Q&A citing his experience as a beta user and presses Marcos on the boundaries of automation without major data science breakthroughs. The dialogue throughout the audience Q&A remains highly collaborative and informative.

Statements from this episode (8)

Disclosure
Mortensen: x.ai is built solely to schedule meetings
“We've set out to do nothing more than schedule meetings. No more, no less.”
Dennis Mortensen Nov 20, 2014 ▶ 0:42
Assertion Not checkable as stated
Dennis Mortensen had 1,019 meetings and 672 reschedules in 2012
“I did a 1019 meetings. And I had 672 reschedules.”
Dennis Mortensen Nov 20, 2014 ▶ 1:13
Disclosure
Mortensen: x.ai features no user interface beyond email CC
“So there is no interface. There is no dropdown box. There is no set of possible times which you can click on. There's only Amy, which exists in dialogue as a CC.”
Dennis Mortensen Nov 20, 2014 ▶ 3:17
Disclosure
Mortensen: x.ai has no backup plan to rely on human labor
“What we kind of set out to do is to kind of create a setting for where there is no plan B. There is no kind of backup plan, and I know SoftBank kind of, you know, shit their pants whenever I say that, but there is no plan B of, ah, perhaps we can just have hum…”
Dennis Mortensen Nov 20, 2014 ▶ 4:39
Disclosure
Jimenez: x.ai uses human AI trainers for QA and data annotation
“Humans right now are, they're not really taking over. They're not really acting as scheduling assistants. What they're acting as is What we call AI trainers. So they're sort of like overlooking what the machine is guessing, so they're doing a lot of, ah, you k…”
Marcos Jimenez Nov 20, 2014 ▶ 6:29
Assertion Not checkable as stated
Jimenez: x.ai correctly identifies 70% of new meeting proposals
“And you can see that we basically, for about 70% of the times that this come in, we nail it, so we know this is a new meeting proposal.”
Marcos Jimenez Nov 20, 2014 ▶ 15:58
Disclosure
Jimenez: x.ai's Amy schedules around learned preferences, not just open slots
“Amy's not just filling up empty buckets in your calendar. She's filling up empty buckets In order to try to reproduce what she deduces are your preferences.”
Marcos Jimenez Nov 20, 2014 ▶ 18:18
Assertion Not checkable as stated
Mortensen: 87 million US knowledge workers schedule 10 billion meetings annually
“So, there's eighty-seven million US knowledge workers today scheduling about ten billion something formal meetings every year, which is a problem you just can't solve with humans.”
Dennis Mortensen Nov 20, 2014 ▶ 24:21
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more