swyx, co-founder of AI Engineer, hosts an open Q&A and feedback session at the AI Engineer World's Fair 2026 in San Francisco. Attendees raise topics including an AI safety track, expo length, overfull workshops, Wi-Fi, and the conference app. swyx discusses speaker selection from 2,200 applications for about 300 spots, plans for future events including a finance-focused New York event and partner AIE conferences worldwide, guidance for running local meetups, and the first academic poster sessions. Chapters cover safety tracks, workshop capacity, staying at Moscone, networking time, app feedback, and speaking selection. # Video Identity Observations: "Building AIE: Open Q&A and Feedback Session with swyx" ## Observable Presentation Overview This video captures an **open Q&A and feedback session** at an AI Engineer (AIE) conference, led by **swyx**, self-described as "one of the organizers of this conference" who is "the guy to answer everything." The format is unscripted audience dialogue — feedback on logistics, programming, culture, and growth strategy. ## Claimed Scope & Positioning - **Stated identity**: AIE is positioned as a professional, industry-facing event, not an academic one. swyx: *"mostly I want people to do stuff that they would use at work. That's like kind of my dividing line."* - **Scale claims**: *"7,000 people"*, *"2,200 applications for effectively 300 spots"*, *"this is only the seventh AIE"*, *"we're three years old"*, and *"I think we're bigger than Shift now."* - **Aspiration to institution status**: *"I want this to be an industry institution. This is not a quick hypergrowth hustle for me... I really believe that AI engineering deserves the top conference."* - **Geographic scope**: Partner conferences in *"Miami, Melbourne, Sydney, Shanghai... Paris... and Singapore"*, plus a TEDx-style licensing model — *"about 40 countries lined up"* and *"the TED versus TEDx model."* - **Community model**: Deliberately no proprietary online community — *"AIE has no community apart from the social media pages and Twitter and LinkedIn and a very inactive Slack... I only focus on in-person community."* ## Intended Audience (as demonstrated) - **Working AI engineers** as the core audience — reinforced by the work-applicability filter for content and the engineering conference app discussion. - **Engineering leaders**, evidenced by a dedicated *"leadership track"* and *"leadership lounge"* discussion. - **Sponsors/expo participants**, with extended discussion of expo staffing and *"world's fairs are meant to be more expo."* - **Meetup organizers globally**, with advice for community-run events (*"meetups are a labor of love"*). - **Students and academics**, via poster sessions and an ACM partnership: *"we want to bring academic academia to industry."* ## Values & Culture Evidence - **Safety posture**: swyx describes himself as *"a little bit of a doomer"* but frames alignment work as emerging from practical engineering: *"through vending bench and through frontier evals... they actually discover way more alignment issues than people who just vaguely talk about X-risk."* - **Anti-corporate culture**: *"I want to be the conference where you would do things that would be unprofessional in the other ones"*; tracks "weird" talks via a CFP checkbox — *"my talk is weird and doesn't fit."* - **Community-vibe index**: a "costume index" (orcs, T-Rex suit, shark) used to measure that the event hasn't been *"taken over by the suits."* - **Transparency practices**: *"we have always open sourced our data"* (MCPS, CLI, JSON files); direct acknowledgment of failures — *"this is something that we failed attendees this year. I'm sorry about that."* - **Iterative feedback loop**: *"We actually really do take the feedback and implement it at our next conference"* — e.g., the NEO (new engineer orientation) *"came out of this kind of feedback session at the last conference."* ## Creator Claims vs. Demonstrated Patterns | Creator claims | Demonstrated in material | |---|---| | Feedback is implemented | Attendees cite improved leadership track; NEO created from prior feedback | | Engineering/practical focus | Rejection of X-risk track in favor of work-applicable content | | Open, weird-friendly culture | CFP "weird talk" checkbox, costume index, Comic-Con analogy | | Not academically accredited by design | *"I like that we are not accredited"* — allows double-presenting | | High speaker bar | Detailed CFP vetting discussion; *"mostly we're known for having a very high bar"* | ## Notable Self-Descriptions - *"My joke is the next step up from here is that I raise money and I build a new building."* - *"We will never leave San Francisco... Moscone is... probably the next five years."* - On next year: *"next year we're not growing... same size, same venue, same everything. And we just optimized."* - *"The business of tech actually just funds real estate."* ## Synthesis-Relevant Notes - The channel/brand is presented as **practitioner-first, deliberately anti-stuffy, and institutionally ambitious** — blending scrappy community energy with large-scale commercial operations. - swyx's persona: candid about failures, opinionated about culture, personally invested in engineering values (mentions *"I am an engineer"*, *"I can vibe code whatever filter I want"*). - The event demonstrates a **dual optimization**: live attendee experience and online YouTube reach (*"YouTube standalone is the format to optimize for"*). # Delivery Observations: swyx — AIE Open Q&A and Feedback Session ## Overall Format - **Audience-driven, improvisational Q&A**: The session has no prepared remarks; swyx opens with, *"If you have questions, we'll do it. If no questions, I'm just going to end early."* Everything else is live responses to feedback. - **Facilitator stance**: He paraphrases audience questions aloud for the room — *"let me let me refra uh let me paraphrase so I can people hear"* — a recurring habit when questions are quiet or complex. ## Recurring Vocabulary and Phrases - **Optimization framing**: He repeatedly describes conference decisions as optimizations: *"I am optimized for that"*, *"I'm kind of just optimizing for getting the best speakers and content"*, *"we basically try to optimize"*. - **"Do stuff that they would use at work"** appears as an explicit decision criterion: *"mostly I want people to do stuff that they would use at work. That's like kind of my dividing line."* - **Vibe coding / open data**: Mentions it multiple times — *"I can vibe code whatever filter I want"*, *"if you wanted to vibe code your own you could"*. - **Contrastive analogies**: He explains concepts through named comparisons — *"the TED versus TEDx model"*, *"It's like Comic-Con"*, *"unconference, right?"*, GDC workshops as a model. ## Teaching and Explanation Habits - **Concrete metrics and lists**: He quantifies constantly — *"2200 applications for effectively 300 spots"*, *"about 40 countries lined up"*, *"this is only the seventh AIE"*, and the playful **"costume index"** (*"two in the first World's Fair, zero in the last year, and then now it's like nine"*). - **Structural breakdowns**: He explains the conference schedule piece by piece — *"Monday, Tuesday, Wednesday, Thursday. The expo only opened at 4 PM on Monday"* — walking through logistics linearly. - **Pragmatic trade-off framing**: Nearly every answer names the trade-off: *"everything is recorded. So, I just say like if if a room is full, go to your number two, go to your number three."* - **Rhetorical self-questioning**: He models his own deliberation aloud: *"what else am I optimizing for? I'm kind of just optimizing for..."*, *"what does that look like, right?"* ## Communication Patterns - **Direct refusals without softening**: *"I don't give it to them"*, *"I don't know if that is my top top priority"*, *"we will never leave San Francisco."* He declines suggestions while acknowledging them. - **Apology and ownership on failures**: *"we failed uh attendees this year. I'm sorry about that"*, *"we promise people that we didn't do it"*, *"I underestimated"* (re: NEO onboarding). - **Deflection via attribution**: When praised, he credits others: *"I had nothing to do with that. I cannot take credit for any of it. It is simply that we are... around longer."* - **Humor as pressure valve**: Frequent light jokes — *"I need them to not have lives"*, *"it feels very 2015"* (quoting feedback), *"the next step up from here is that I raise money and I build a new building."* - **Invitation to continue criticism**: *"Do you want me to keep going like there's a—" / "Yes, I do. Uh because all this is going straight to my team."* ## Recurring Themes - **Repurposing feedback into commitments**: *"We actually really do take the feedback and implement it at our next conference"*; NEO is cited as *"came out of this kind of feedback session at the last conference."* - **Longevity over growth**: *"I want this to last decades. I want this to be an industry institution. This is not a quick hyperrowth hustle for me."* - **Place for the weird**: *"my talk is weird and doesn't fit and I always want to be have a place for weird stuff."* ## Uncertainty Notes - The session is a single feedback format; his **lecture-style delivery** (e.g., his keynote) is only described, not observed, so teaching habits here reflect facilitation rather than presentation. - The "costume index" joke appears once; whether it is a stable running gag cannot be confirmed from this transcript alone. # Position Record: "Building AIE: Open Q&A and Feedback Session with swyx" ## Speaker Identification - **swyx (Swix)**: Identifies himself at the start as "one of the organizers of this conference"; the session is an open Q&A and feedback session. Multiple unnamed audience members ask questions; a few identify themselves by name (Eric Nes — Minneapolis AI engineer meetup organizer; Justin; Alex; an attendee from Bangalore; an attendee representing a nonprofit sports-industry organization). - Where the transcript uses `>>` without a name, the speaker is identified only by conversational context (questioner vs. swyx's replies) and is recorded below as "audience member" where identification is uncertain. --- ## Conference Logistics & Announcements (swyx) - **Future conferences**: Announces AIE will return "a year from now in 2027" and in **New York**, "focused on finance and anything money and AI." - **Size freeze**: States (as a "special preview" before the closing keynotes) that next year will **not grow** — "same size, same venue, same everything" — so the team can optimize rather than scale. - **Venue commitment**: "We will never leave San Francisco. Moscone is it at least for this year and next year and probably the next five years." Jokes that "the next step up from here is that I raise money and I build a new building." - **Long-term framing**: "I want this to last decades. I want this to be an industry institution. This is not a quick hypergrowth hustle for me." - **Partner conferences**: Lists four to five international partner events — Miami, Melbourne, Sydney, Shanghai (TBD, targeted November, has "shifted back two times already"), Paris, Singapore. Growth is "only by geography," not venue size. - **Personal note**: He has lost his passport and is "stuck in the US" until replaced. --- ## AI Safety / Existential Risk Track Debate This exchange contains a visible disagreement that should be preserved, not smoothed. - **Audience member** asks about an AI safety / existential-risk track next year. - **swyx's position**: - X-risk content is "a form of like discussion and research and theory," which he does not put as a track because "mostly I want people to do stuff that they would use at work. That's like kind of my dividing line." - Discloses a qualification: "I'm familiar with it. I'm a little bit of a doomer. I have I've said this before in my public statements." Immediately follows with: "it is just not the purpose of an AI engineer to help with that." - Offers alternatives: attendees can self-organize meetups (cites an AI for science meetup as precedent). - **Audience pushback**: The questioner responds that "it should be the job of the AI engineers as much as anyone else to be considering" the "endgame" of the loops being built. - **swyx's partial concession / reframing**: - Points to existing safety-adjacent content: a security track, sandboxing track, guardrails, evals, and an eval track with labs. - Claims frontier evals "actually discover way more alignment issues than people who just vaguely talk about X-risk." - Argues engineering is itself alignment: "the statement that engineers can wield AI and control AI in the systems is fundamentally a control structure over AI anyway. So everyone here is like philosophically aligned to it without necessarily having to put it front and center." --- ## Expo & Conference Format Feedback - **Audience member** suggests the expo (four days) is too long and proposes: one dedicated partner/networking day, one conference/workshops-only day, expo for two days. - **swyx's response**: - Notes he also "had complaints that there was not enough expo" (expo opened at 4 PM Monday; people asked why not the full day). - "As long as people staff the expo... and sponsors pay for it, I'm very happy keeping it open." - Distinguishes formats: **summits** are "very curated and focused on content"; **world's fairs** are expo-oriented, where "you can get an expo ticket and walk around all day and get a sense of the field." - Admits he has considered "even more expo upstairs" (a content expo) because he dislikes "the time pressure of talks," but concludes "I don't know what the right mix is." - Defends the music corner against complaints: "I like music." - **Tension preserved**: swyx acknowledges the "content max vs. free time" trade-off as "an open debate" but says the option not to attend sessions is always available and he optimizes for speakers/content going online. --- ## Overwhelming Schedule / FOMO - **Audience member** (joined late; self-identifies as having attended many of these): reports feeling overwhelmed — "so many talks... at some point I kind of just gave up," losing the ability to talk to speakers afterward; suggests fewer but higher-quality talks. - **swyx's response**: - Names it "anxiety, right? FOMO" and says it's something he "do[es] want to manage," but calls it "a high class problem... an embarrassment of riches." - Points to **summits** (single-track) as the alternative format; "the world's fair is the big show." - Argues discomfort is appropriate: "we're covering the fastest moving fields ever. And so, you should be a little bit uncomfortable... if you're completely comfortable, you're probably not being pushed." --- ## Workshops & Registration Failures - **Audience member** reports spending "half the day waiting in line" for workshops without getting in; asks for online pre-registration. - **swyx's response**: - Owns the failure: "we failed attendees this year. I'm sorry about that." - Says they "paid a vendor a lot of money for the mobile app" and pre-registration "wasn't the right fit"; he personally "wasn't involved." - Notes the trade-off: "everything is recorded," so overflow attendees can go to their number two or three choice; his optimization is "the UX of who gets in the room and then how quickly do you know you're not getting in." - Acknowledges: "people always want registration... I register for things and I also don't show up." - **Workshop length feedback**: An audience member found 20-minute workshops too short for substance; praised full-hour workshops. **swyx** says lengths *were* mixed (45-minute and longer rooms existed) but concedes "nothing I went to was long... that's on that's on me." Cites **GDC** as an inspiration; argues online attention spans favor 20-minute condensation; states "I do want people to go deep. Many people don't deserve the time, but they want the time, and so I don't give it to them." --- ## Mobile App & Event Technology Multiple attendees provide detailed app criticism; swyx responds at length. - **Audience complaints** (recorded as reported): - Android crashes on re-entry; slow loading screens when navigating back; scroll position lost or loading delays when scrolling. - "My schedule" tab lacks search/filter, and forces scrolling from the first day of the event. - Cannot save currently-ongoing sessions. - No sync between web and app (different code bases). - Booth listings lack names/locations; unused chat/Q&A features; unoptimized filters; no location filter. - One attendee calls the app "a slob"; swyx agrees "it's rough." - **swyx's positions**: - "If I have my way, that will go away next year and we will just do the web app"; "we want to get rid of that vendor." - Has long resisted making an app because "they all suck"; notes AIE has **always open-sourced its data** (JSON, MCPs, CLI) so attendees could vibe-code their own apps — "if there's any conference where you can vibe code your own app it's this one." - Wants to "take control" of the app next year, now that "I finally understand what we need to show people... it took me three years... this is only the seventh AIE." - Frames the recorded feedback session as the "requirements document" for next year's build; asserts AIE is building capability in **event technology** as a discipline separate from event organization. - Mentions a leadership-track matching prototype ("we built a Facebook where we... deep research every leadership attendee") and wants to invest more in matching, having "never seen the chat work" at any conference. --- ## International Expansion & Community Model - **Attendee from Bangalore** (second AIE world's fair) asks about India expansion and local chapters. - **swyx's positions**: - "You're the fifth Indian to ask me that"; acknowledges high demand. - A partnerships program ("about 40 countries lined up") using a **TED/TEDx model**: "I run the main TED and everyone can have the TEDx in their country... I want the brand to still stand for something... but I don't want to be the limiting factor." Bangalore has an interested partner he's been talking to since AI Singapore. - **Community philosophy**: "AIE has no community apart from the social media pages and Twitter and LinkedIn and a very inactive Slack." Prefers in-person community and living on existing platforms: "I don't want to add yet another Slack to your already 200 Slacks." Notes attendee tweets are "free marketing for AIE." - **Justin** asks about the Shanghai waitlist (TBD status); swyx confirms and explains the rationale: "we want to bring American researchers and engineers to meet their Chinese counterparts and have that exchange... we should do the other way too." - **Eric Nes** (Minneapolis meetup organizer) asks about production tips for local meetups. **swyx advises**: - Meetups are "a labor of love"; use volunteers, never do it alone. - Have a consistent home ("every fourth Thursday"). - Recording/polish is optional; meetups need not be mini-conferences (cites Portland beer-garden meetups). - Points to **js.la** as a model: "if you figure out the event organization and then publishing, that's the 80/20 of it." - Warns: running a full conference "can end up being your full-time job... it's not for everybody." --- ## Leadership Track & Unconference Ideas - **Alex** (first-time attendee) praises the leadership track but asks for more free time for leaders to share experience (open-ended sessions on org, product, design). - **swyx**: Agrees but "don't know if that is my top top priority"; cites the leadership lounge and attendees' autonomy. Notes **London summit leadership lunches** (5–10 min guided discussion, then hour of table chat around three primed topics) were "very well reviewed" and could return at summits. - **Nonprofit attendee** (sports industry, four years attending, spends "tens of thousands of dollars" bringing the team) praises the leadership track as "significantly better this year" — less "salesy or hungry for relevance" — and describes their pre-conference internal prep and nightly hack sessions. - **swyx's candid qualification**: "I had nothing to do with that. I cannot take credit for any of it... It's all inbound... people stepping up saying I have something to share. Because we've been around enough that... we're established now." - **Table topics / unconference**: Audience member recalls year-one "birds of a feather" table topics and suggests revival. swyx calls it "unconference" and expresses doubt: "people want us to do an unconference. The problem is I worry that we don't sell tickets... I wouldn't go if that was it." Hints at a possible dedicated **leadership retreat** (Napa weekend) where unconference and off-the-record chats could fit. - swyx also floats converting upstairs space into a "content expo" (see Expo section). --- ## Conference Culture & "Suits vs. Hoodies" - **Nonprofit attendee** worries the event will lose its scrappy vibe as it succeeds (South by Southwest analogy: "the ratio of sport coats to hoodies started to take over"). - **swyx's response**: - Disagrees it has happened yet, citing his **"costume index"**: two orcs, a T-Rex suit, ducks, a shark; the index "was two in the first World's Fair, zero in the last year, and then now it's like nine. So I think we're going in the right direction." - Aspiration: "I want to be the conference where you would do things that would be unprofessional in the other ones... you can be yourself." Compares to **Comic-Con** as a culture that survived success. - "We do still want to serve the suits... there's a floor of the Moscone separating the suits from here." - Institutionalizes weirdness in the CFP: the last track option is "my talk is weird and doesn't fit" — "I always want to have a place for weird stuff" (cites Eric Meyer's and John AuYeung's talks). --- ## Speaker Selection (CFP) Feedback - **Audience member** (accepted for remote track; submitted five applications for his spatial-reasoning/architecture team, Series C, novel computer vision) asks how non-Twitter-famous, non-Silicon-Valley applicants can improve selection odds. - **swyx's explanation of the selection process**: - "~2200 applications for effectively 300 spots," many of which go to invited speakers; "it's very very competitive." - Helps to be "somewhat frontier," unique, and tied "to a broader trend that we can tell." - Admits trade-offs: some vision-track slots go to proprietary-model holders and "leading companies that are somewhat Twitter famous because that puts butts in seats"; "then we have some leftover for... let's see how they do." - Applications are **not blind** by design; prior recorded talks are watched — "if we go this guy is really goddamn good, we'll fight for you." - Self-critical note: "It's a lossy bar. Like some people get through that should not get through... but mostly we're known for having a very high bar." --- ## Academia & Student Access - **Audience member** thanks swyx for student/academic access via the ACM agents partnership and advocates for more young researchers. - **swyx's positions**: - AIE "was started kind of in opposition to NeurIPS" — e.g., not cutting submissions six months out ("that guarantees everything's out of date") — but aims to "bring academia to industry... in the same way that NeurIPS brings industry to academia." - ACM/"Kais" was "the easiest to work with by far"; selection was fully outsourced to them; eventually "we're going to have a peer review process" and possibly "our own journal." - Values non-accreditation: presenters face no exclusivity rules and can "double present." - Notes employer interest in hiring students; "Worlds Fair is basically the only place I can do this because I have the real estate." --- ## Track Continuity & Content Curation - **Audience member** asks how a four-hour block maintains a common thread given repeated subject matter. - **swyx**: Manually sequences and clusters talks ("I know the content of every single talk"), but scheduling is constrained by speakers' availability ("I need them to not have lives... but this is just not realistic"). Agrees "YouTube standalone is the format to optimize for" and maintains per-track playlists "that nobody uses." - **New Engineer Orientation (NEO) feedback**: An attendee found it under-thorough (lunch, schedule basics). swyx responds that NEO "came out of this kind of feedback session at the last conference" and went "from zero orientation to a little bit"; his team underestimated the load; an attendee guide existed but "I don't think most people saw it." Ties improvement to the size freeze ("we have a year to prepare"). --- ## Wi-Fi Complaints - **Audience member**: Wi-Fi was bad; workshop presenters couldn't run demos. - **swyx's response**: - "We got complaints, we got demands for refunds, everything. It happens every year"; they "pay maximum for whatever the Wi-Fi package" the venue offers. - Structural diagnosis: venues "are not set up for engineers downloading gigabytes of things per laptop." - Improvement claim: better than "basement 3 of the Marriott" last year — "here, at least everyone can tether." - Separate **speaker Wi-Fi** and **sponsor Wi-Fi** existed ("I'm not sure that every speaker was told about it"). - Concession: "It's never been great at any conference I've ever been to. So that's the uncomfortable answer." --- ## swyx's Keynote & Closing Notes - References his Tuesday keynote: "everyone is doing loops," demonstrated via a CLI auto-touring the talks ("only Opus decided to do an auto tour"). Declines demands for more talks: "I have a podcast, I don't need to speak." - On AIE's origins: "AIE was started inspired by Shift in Croatia. Shift is one of the best conferences I've ever been to"; Ivonne (Shift) is "a mentor of mine for organizing AIE... and potentially running it as a business that can be sold." - Notes AIE is "three years old" vs. Shift's 15 and claims "we're bigger than Shift now"; says he is "sensitive to how we fit in the broader scale because we're not the only conference that people will go to." - Shout-out to Yan as recurring mic-holder and multi-time speaker. - Closes by heading to "the startup battlefield" and inviting attendees back next year. --- ## Unresolved Tensions (Preserved) 1. **Safety track**: swyx calls himself "a little bit of a doomer" yet says X-risk "is just not the purpose of an AI engineer"; the questioner's pushback that engineers *should* consider endgames is met with reframing, not a commitment to a track. 2. **Content density vs. breathing room**: swyx explicitly labels the "content max or schedule nothing" question "an open debate" and resolves it in favor of content plus attendee autonomy. 3. **Expo length**: Simultaneously "too much expo" and "not enough expo" complaints were received; swyx offers no resolution beyond "I'm very happy keeping it open." 4. **Speaker selection**: Explicitly balances frontier/uniqueness against Twitter fame "putting butts in seats" — acknowledged as lossy, with no fix proposed beyond applicants submitting stronger proof of prior talks. # Speaker Roles & Provenance Record ## Video Metadata - **Title:** Building AIE: Open Q&A and Feedback Session with swyx - **Format:** Live open-ended Q&A/feedback session at an AI Engineer (AIE) World's Fair conference ## Identified Speakers and Roles ### Primary Speaker - **swyx (Swix)** — *Conference organizer/creator.* Self-identified: "I'm Swix. I'm one of the organizers of this conference." Speaks in the capacity of organizer answering feedback, not as an independent content creator in this session. Contact given as **swix@ai.engineer**. ### Audience Questioners (names when self-identified) - **Unnamed questioner(s)** — raised AI safety/x-risk track, expo length, workshop registration, venue size, Wi-Fi, mobile app feedback. - **Eric Nes** — organizer of the **AI Engineer meetup in Minneapolis**; asked about production value for local meetups. - **Attendee from Bangalore, India** — asked about AIE chapters/expansion to India. - **Justin** — first-time attendee; asked about Shanghai event timing; later gave detailed mobile app bug feedback. - **Alex** — first-time attendee; feedback on leadership track free time. - **Yan** — mic runner; described by swyx as a "multiple time speaker." - **Anonymous attendee (Kevin/Andy referenced)** — feedback on talk overload/FOMO. - **Attendee from a nonprofit in the sports industry** — 4-year returning attendee; praised leadership track. - **Attendee who applied via CFP** — spatial reasoning/architecture company (Series C); asked about speaker selection. - **Attendee via ACM agents project** — praised student/academic access. ### Named Third Parties Mentioned - **Ivonne** — mentor to swyx, associated with **Shift conference (Croatia)**. - **Jazelle** — AIE team member referenced re: MCP block sequencing. - **Eve** — head of design for **Y Combinator**, spoke on the design track. - **Eric Meyer** and **John Auster** — speakers from the current event. - **Sharina Grim** — AIE Singapore team; built the networking app. ## Referenced Sources and Third-Party Claims - **Conferences referenced:** Shift (Croatia, ~15 years old), GDC (Game DevCon), TED/TEDx, NeurIPS, ICML, re:Invent (AWS), South by Southwest, We Are Developers, Comic-Con, ACM agents conference. - **Partner conferences/locations claimed:** Miami, Melbourne, Sydney, Shanghai (TBD, targeted November), Paris, Singapore; ~40 countries in partnerships program; partner **Gortech** mentioned. - **Claims by swyx (recorded, not verified):** ~2,200 CFP applications for ~300 spots; this is the **7th AIE**; ~7,000 attendees; conference started ~3 years ago; inspired by Shift; data open-sourced (JSON files, MCPs, CLI); AIE described as "very safety aligned." - **Vendors referenced:** Paid external vendor for the mobile app (slated for removal next year); external vendor for chat/Q&A features; workshop pre-registration attempt with vendor failed. ## Availability / Access Facts - All sessions recorded; stated intention to publish to **YouTube** with per-track playlists. - Attendee guide exists online (swyx notes most people didn't see it). - Conference data is open-sourced for attendees to build their own apps. - No statement in the transcript about comment availability on any platform. ## Promotional Segments - swyx repeatedly **promotes AIE itself**: future 2027 return, New York (finance-focused) conference, expansion plans, partnerships program, and the open-data initiative. - Mentions of **Shift (Croatia)** and **Kais/ACM partnership** function as endorsements rather than paid promotion. ## Uncertainty Flags - Several audience questioners are unnamed in the transcript; attribution to specific individuals is limited to those who self-identified. - Speaker labels (`>>`) do not distinguish all voices; some interleaved dialogue attribution is approximate. - Transcript contains transcription artifacts (e.g., "Swix/Swyx," "Mosone/Moscone," "Ligh Haven"); names rendered as heard. Hi everyone. Uh I'm Swix. I'm one of the organizers of this conference. Um basically this is one of these open-ended Q&A feedback sessions. If you guys have feedback or Q&A about how any of this conference is run, uh I'm the guy to answer everything. We actually really do take the feedback and implement it at our next conference. Uh we're going to announce that we're back here a year from now in 2027 as well as in New York uh which is our next conference immediately focused on finance and anything money and AI. Um so this kind of very open-ended. If you have questions, we'll do it. If no questions, I'm just going to end early. So, over to you. Anyone has questions? Yeah. \>\> So, uh let me let me refra uh let me paraphrase so I can people hear. Uh what about an AI safety or existential risk track next year? Um that usually happens in Ligh Haven up north in Berkeley. Um I would say that uh typically it is not so much engineering. Um I would say that typically it is a form of like discussion and research and theory that uh we typically don't put as a track. Uh because mostly I want people to do stuff that they would use at work. That's like kind of my dividing line. Um that said, we're very pro people organizing their own uh meetups. We have places to for uh attendees to present uh which I think is very important for people to speak. Um we'll have other venues like uh for example we also did not have AI for science this year and obviously science is a thing that a lot of people do for work. Uh but people just organized meetups. There was an AI for science meetup and there was there's there can be a meetup or X-risk. Um I'm familiar with it. I'm a little bit of a doomer. I have I've said this before in my public statements. Uh, it is just not the purpose of an AI engineer to help with that. \>\> Yep. \[laughter\] \>\> Hello. \>\> Sorry. Okay, we're on. \>\> Um, I guess my only response is that I almost think it should be the job of the AI engineers as much as anyone else to be considering this sort of like \>\> ongoing loop machine that we're now talking about building and sort of considering okay, what is the endg game of that? But yeah, thank you. Yeah to to totally and obviously it's not something that I'm people are aware of it there's a whole security track there's a sandboxing track uh you know there there's this appears in a lot of our work like if you have any guardrails if you have any evals um our eval track open with labs we're actually a very safety aligned company but they they do vending bench um and through vending bench and through frontier evals that involve the real world they actually discover way more alignment issues than people who just vaguely talk about X- risk, right? Like so to me it's like very interesting to directly point towards a thing that people do at work and the thing that is like leading in the frontier and along the way solve alignment or solve X- risk. Um I think at the end of the day the statement that engineers uh AI can be engineered or engineers can wield AI and control AI in the systems is fundamentally a control structure over AI anyway. Um, so everyone here is like philosophically aligned to it without necessarily having to put it front and center. \>\> Uh, first of all, first of all, uh, great event. Uh, and I have not so much of a question as, uh, maybe a bit of feedback. Do you think, uh, the expo, the expo itself, uh, like maybe four days is too much? \>\> No. Uh, do you think so? \>\> Yeah. Yeah. Yeah. I think it's uh kind of pretty like it was kind of half that today almost. I would also like think maybe just have a one just fully dedicated day to just like partners on like some networking events and just like session and so maybe one day like conference style and workshops only and maybe have the expo only for two days u I mean you know \>\> yeah okay um I'm not sure that everyone would agree \>\> um and that's the first thing the second thing is actually I had complaints that there was not enough expo \[laughter\] Uh, so the the structure of the conference is Monday, Tuesday, Wednesday, Thursday. The expo only opened at 4 PM on Monday. And I had people asked me how come it wasn't open on Monday the whole day, you know. Um, all I will say is as long as people staff the expo and and stuff happens and sponsors pay for it, I'm very happy keeping it open. Uh I actually think world's fair is different from my other summits in a sense that summits are meant to be very curated and focused on content. The world's fairs are meant to be uh more expo actually like this is the the kind of expo that I always imagined where you can just get an expo ticket and walk around all day and get a sense of the field. Uh and if you can't make Monday you can come to Tuesday or Wednesday or Thursday which actually people do. People come on the single days only because obviously they have work. Um other things about this um I actually what's very funny is I actually have thought about what if there was even more expo upstairs. Uh because I I don't like the time pressure of talks. Um I think that you know having to rush to this room that room that this room is very pressuring on everybody attendees and speakers and even us as organizers. The natural way that people have conversations here is just kind of walking around I think is great. So, what if we turned upstairs into like kind of a content expo? Here is like the sponsor expo and people just stood around. It's also very tiring to stand around. So, I don't know what the right mix is. Um, what else would I say about this? We also in insert a lot of breaks like you know, we spent a lot on like the puppy corner and like the uh the poster sessions and the music corner and like you know, Grav even brought in the uh you like the music corner. Um, people complained about the music corner. They were like, "Ah, it's like interrupting my quiet time outside." And I'm like, "Well, you can go upstairs to the leadership lounge. You can come downstairs to the expo." I don't know. Um, I like music, so I at the music \[laughter\] corner. Um, \>\> uh, can we have you as a speaker back at Shift in Croatia? \>\> Oo, I love Shift. I love Shift. \>\> You're coming back this year or next? \>\> Uh, I have just lost my passport. I'm stuck in the US until I get my new passport. Um, but yeah, so uh for those who don't know, AIE was started inspired by Shift in Croatia. Shift is one of the best conferences I've ever been to and spoken at. Um, and Ivonne is, you know, a mentor of mine for organizing AIE. Uh, and potentially, you know, running it as a as a business that that can be sold. Um, yeah, I would love to come back. I just haven't really I I you know now I now I travel to other AIES around the world uh by uh you know this this year basically we're between four to five partner conferences in Miami, Melbourne, Sydney, Shanghai uh Shanghai is still TBD but uh Paris is also going to happen uh and Singapore and so I already have a lot of travel then like adding travel to nonE conferences is is is a big ask but shift is We also had uh visitors from we are developers and other some other uh large conferences. Uh we're apparently starting to show up on other people's radars you know like we are a lot younger. I don't know how old is shift like seven eight years old \>\> 15. Yeah. So like we are three years old. Um and bigger than a lot of other I think we're bigger than Shift now. um like it's it's it's interesting that like how we merge in with the other conference um companies and and how we figure in the landscape of the marketing charts and time calendars of of uh you know our sponsors and our attendees uh and so I'm I'm kind of sensitive to how we fit in the broader scale because we're not the only conference that people will go to. \>\> Hello. Hi. Hi. So I'm coming down from Bangalore. uh this is my second AI engineers worldfare here uh that I'm attending. So I was just wondering that how AIE chapters are growing. Is there a plan for bringing this to India? Probably Bangalore or any of uh the cities eventually. \>\> Yeah. You're the fifth Indian to ask me that in this. \>\> Exactly. We need that. \>\> Yeah. Yeah. Yeah. No, obviously there's high demand and there's a lot of people in India. Uh yeah, there is uh we have a partner uh Gortech. I think uh you know them. Um basically we have partnerships program. They're on the website. You go to AI.engineers. engineers scroll down you see the partnerships section we have about 40 countries lined up that all want to do some kind of AIE in their country and we are not going to reach every country right so uh the the idea is just like to scalably do the TED TED versus TEDex model of like okay I run the main TED and everyone can have the TED in TEDex in their country as long as it serves them I don't care like you know like I mean actually I I care in sense of like I want the brand to still stand for something uh and have meaningful conversations and and have good technical content Um but I don't want to be the limiting factor on countries having their own community and if AI can help that then why not u and so so India is one of those places where I'm like I don't really know the country that well but if you guys want to take it and run with it yeah absolutely sure so that we have somebody that I've been talking to since um AI Singapore that wants to do it for Bangalore um and presumably the other cities as well I think we probably start with Bangalore. Yeah. No, that that makes sense. And is there a plan for like communities like these the chapters that like uh let's say people can join and we can join and be a part of it? \>\> Yeah. So actually AIE has no community apart from the social media pages and Twitter and and LinkedIn and a very inactive Slack. Uh it we you know I only focus on inerson community. I think it would be very interesting to build online community, but I much prefer us live on other existing platforms because that's where people already hang out. I don't want to add yet another Slack to your already 200 Slacks. I have too many Slacks. Like, I don't need another Slack, but I'm already in Twitter. AIE is integrated in Twitter. By the way, it's free marketing for AIE every time you tweet about AIE. Like, it's there's just many reasons to just integrate on your existing social media. anyway your existing blogs um and we we get by on that uh we there are meetups um AIE Italy has been meeting for a long time a London has been meeting for a long time u and I think there's I think AIE Argentina there is one of the South American countries is meeting so people can organize meetups we provide support um and uh I think that's that's the community really like I would don't even think about it as the AIE community I think about it just more as like if you're an engineer in AI you will sometimes come to AIE Sometimes you go to other not as good conferences and then you think about how great AI was and you come back. \[laughter\] \>\> Awesome. Thank you. \>\> Oh, thanks Yan for being the the mic guy. \[laughter\] Yan is a like multiple time speaker as well and like uh it's great to have you here. \>\> Hey. Hey. It's uh Eric Nes. I organized the AI engineer meetup in Minneapolis. \>\> Hey. \>\> Hey. \>\> We were just talking about this. \>\> Yeah, you can start your own meetup. You just \>\> you can just do things. \>\> DM Swix. \>\> Just DM Swix. Uh, so the production here has been awesome and I'm just wondering like for local meetups, do you have any quick tips for how to up production value? production value. That's an interesting one. Um, I don't because I think it is or I organized meetups for years before I did my first conference talk and I think that meetups are a labor of love and they're always last minute and you're always looking for speakers, you're always looking for venues, you're always ordering pizzas and what have you and that's that's the norm. Um the production it usually just involves a group of volunteers that all help each other out. Like you don't definitely don't do it alone. I think definitely having a home where like people consistently know this is the same venue, same time of the month like let's say we meet every fourth Thursday or something, you know. \>\> Yep. Um, and then lastly, it's just always the polish in like selection of speakers and then the way you record it, if you record it. Um, I definitely do not think you need to record the whole thing. Basically, I think there are some meetups that run effectively as mini conferences where they do two to five talks a meet every meetup and it's just talk talk talk and then c this discussion and then you end. I think that the the possible things that you can do at meetups are much wider that you can do other activities that mingle a lot more and you can um in Portland they used to do these like beer things and you just go out and have beers in a beer garden. Um and that's fun. That's a meetup. Uh it doesn't always have to be like a mini conference talk like this like this stuff. \>\> Yeah. We usually do like some social time, two talks and then a little bit more social time. So \>\> yeah. So I would say that that is the norm and that is uh what a an average meetup should be \>\> and then and then if you want to add a little bit more flare on it, you have the website, you have the prestigious company sponsoring it, you have like top speakers whenever they're in town, they go they go to that meetup because that's the one that's known because it has a Twitter account and a YouTube channel and all has all this stuff. It's just all extra work on top of just organizing the meetup. \>\> Okay. \>\> Um maybe you can talk to JS uh look at look for JS.la LA I would say as that's a a pretty successful meetup. \>\> Okay. \>\> Uh that is wellrun well produced. Uh it like I think if you figure out the event organization and then publishing that's the 8020 of it. You don't really need a ton else of like infrastructure or company organization around it. \>\> Um \>\> and to some extent I I want to keep it the bar low and accessible. You can always up it. You can always start end up running your own conference. Uh just be prepared for it to be your full-time job \[laughter\] and uh it can it's not for everybody. \>\> Okay. \>\> Yeah. \>\> Well, thank you. I' I've really enjoyed myself here. \>\> Thanks for coming, man. \>\> Thank you. Yeah. \>\> Uh more questions, more feedback. Um again, yeah. \>\> Uh so mainly my feedback is around like I think Monday the workshops, right? I feel like if there's a way to like sign up online for a room because like I spent half the day waiting in line to get into a room that I never got into. \>\> Yeah. Um yeah, this is something that we failed uh attendees this year. I'm sorry about that. Um some some like we have roughly equalized rooms, but we don't have equaliz allocation to make. I mean the trade-off is that everything is recorded. So, I just say like if if a room is full, go to your number two, go to your number three. Um uh but yes, people want to pre-register. Um we tried to we paid a vendor a lot of money for the mobile app for the conference and we couldn't make it work for the pre-registration for the workshops. I don't I personally wasn't involved. So, my team just said like we tried it and it's not the right fit. So, next year we'll try it. Uh people always want registration. Uh I will I will just generally say that um you know I register for things and I also don't show up. So \[laughter\] uh I think there needs to be some better organization that that we owe you as an attendee to do that we haven't done and you know I I'm disappointed in that we didn't do it this year. I I I seriously thought this was the year that we would start doing workshops uh registrations. Um and we promise people that we didn't do it. Um it's mostly a software issue. We got to figure it out. Thank you. Again, it's all recorded, right? Like, you know, I I think that it's the UX of who gets in the room and then how quickly do you know you're not getting in the room so you can go to the next thing. So, uh I am optimized for that. I know that that is a pain point. Um yeah, I'm I heard as they say maybe I I have one as well. Yeah. \>\> Uh is the maximum size of the conference that you're planning? I mean given the capacity of the venue because I I feel like it's it's not necessarily like the bigger the better always. \>\> Yeah. Yeah. I agree and um we did go for size in uh as far as the world fair is concerned because size has its own benefits because it can pull some of the top speakers that we have. Um but then there's also such a thing as too big. So I haven't announced this yet but you guys get a special preview. We're going to go on to the closing keynotes and say next year we're not growing. So next year, same size, same venue, same everything. And we just optimized. Uh it was the idea that now all of our sponsors have seen what a world's fair at Moscone looks like. All of our attendees have seen it. All of our um speakers have seen it. That's all. And most importantly, my employees have all seen it. So now we actually know what a world's fair looks like at Mosone. We're just going to do this every year until we get good and then we can grow. And so I think probably next year like I there was some rooms that were still empty like level three was still kind of I think we can just optimize the usage of the existing space better but we have the exact same week booked next year and so we can just come back like we you know we I so I'm very comfortable making the promise of that nothing like we won't grow in terms of like changing to a larger venue or anything like that and I think I want everyone to have a much experience including us organizing so that um you know this becomes a sustainable industry conference like I want this to last decades. I want this to be an industry institution. This is not a quick hyperrowth hustle for me. You know I I really believe that AI engineering deserves uh you know the top conference that uh that we can kind we can run. So yeah, I I I feel the the the the the the comments mostly from Europeans. Uh which is like very interesting. Um but yeah, I I I definitely don't like the very very very very big like reinvent type AWS things in in Las Vegas. So we will never leave San Francisco. Um Moscone is it at least for this year and next year and probably the next five years. Um, in fact, my joke is the next step up from here is that I raise money and I build a new building. Uh, because I there's there's things about Moscone I don't like. So, um, what does that look like, right? Um, so we we'll see. Uh, but you know, I think I always joke like the business of tech actually just funds real estate. It's not the, you know, it's not the other way around. Anyway, I I that's a little bit of a ramble, but um there is that that is the intention that we only grow by geog geography and like accessibility to other countries, but as far as San Francisco is concerned, this is we're going to stay put for a while. \>\> Thank you. \>\> I don't know if this came up, but the Wi-Fi really was bad. Like even the presenters were not able to present their demos in the workshops mainly. I don't know if it's up to you or \>\> No, we got we got uh we got complaints, we got demands for refunds, everything. Uh it happens every year. Um uh and you know, we we pay maximum for whatever the Wi-Fi package they offer us is. \>\> And so beyond that, what can we do? Um and the the truth of it is most of these venues are not set up for engineers downloading gigabytes of things per laptop every time. Like they they're not that kind of conference venue. Um, so, uh, to me the this is an improvement from last year because we're no longer in basement 3 of the Marriott where there's no reception. Here, at least everyone can tether. Uh, and that's good. As far as speakers were concerned, uh, we actually did have a separate speaker Wi-Fi. Uh, I'm not sure that every speaker was told about it, but as far as I I was always on the speaker Wi-Fi and I never got any issue. Uh, and also sponsors have a separate Wi-Fi as well, so we want to make sure the sponsors can do their demos. Um but as far as attendees, yes, there will be congestion. Um we try to pay for the max of everything. The other, you know, I think the next option from here is I don't know build buying our own hots hotspots and like giving everyone like second Wi-Fi things. Um it's not great. It's never been great at any conference I've ever been to. So I that's that's the uncomfortable answer that I have to give right now. Unless you know a Wi-Fi person. \>\> Yeah. How many satellites do you need? How many dishes do you need to serve 7,000 people? I don't know. Starting cool though. \>\> Hey, how's it going? Uh my name is Justin. This is my first uh AI engineer. \>\> Welcome event. Um I frequently travel to China. Uh I saw that there's a Shanghai uh possibly but you just mentioned that it's TVD. \>\> Yep. \>\> So is that why uh there is a weight list? \>\> Yep. \>\> Okay. Okay. Just wanted to confirm that and make sure. \>\> Uh thank you. \>\> Yeah. Uh right now targeted at November. This date has shifted back two times already. Um \>\> I don't know we uh we we these are all we rely on external partners uh that we work with to organize the international events. Uh but we want to do Shanghai. We want to bring American researchers and engineers to meet their Chinese counterparts and have that exchange because I think uh that will at least give people an opportunity to understand who's on the other side. Um I mean it's pretty good that they come here but we should do the other the other way too. I I I find it interesting. That's your first That's your first question. You given that it's your first time coming to AIE. You didn't have anything else. You just want this to be in Shanghai. That's your question. \[laughter\] Uh okay. Other questions or feedback? We don't have to take the full period. \>\> Yes. Oh, hello. \>\> Hey. Um, my name is Alex. So, first of all, thank you for such a great event and uh I'm first time on AI engineer and it's really cool. Um, I really like the idea of the leadership track and really cool stuff there. The uh uh the one improvement there that I I think would be really good which we actually experienced earlier this morning is to have more like free time to chat and share the experience among the leaders. So you you have a lot of great speakers already lined up but with all the leadership you also bring a lot of experience across the a lot of companies having the opportunity to just share the experience with each other and maybe have open-ended sessions with the topic saying like how the organization work how you figuring out the product how you figuring out the design. I think that would be a great addition to this like just let people to share their experience. \>\> Yeah. Um, I agree, but I don't know if that is my top top priority because people can still exchange contacts and meet tomorrow. Tomorrow is Friday. It's a holiday. People can still meet. They don't have to meet here. Um, they also there was also the leadership lounge where people could hang out all day. This is basically like if you want to chat, you can just not go to the sessions. So like it's it's that that is that is an open debate, right? Right. So should I content max or should I use the time and just schedule nothing in that time and just let people talk or just give you an option. \>\> That that is an that is an option. So to me the option to not attend sessions is built into everything at AIE, right? You can just not attend anything and just talk and a lot of people do that. And so then a question of like okay I've I think you have your own free will and autonomy to do all those things. You have the space the the lounge um to do that. Then what else am I optimizing for? I'm kind of just optimizing for getting the best speakers and content I can online so that everyone can benefit online. And so then I just give them the space. And sometimes people speak to rooms where it's like 80% not full. it's okay because they get their stuff, they get the story out. Sometimes it's very very good and actually it does a lot better online than it does in person. Um, so then I'm like, okay, I much rather have that than leave it blank where we pay for this space and we didn't use it. But I understand this this perspective. I would say that um as far as the summits go or like in in London for example, um there was no separate leadership track. There was basically leadership lunchons and the lunchons were very good because we had basically five 10 minutes guided discussion and then the the rest the rest of the hour was just chat across tables among everything and we focused the conversation around like three topics that we that we sort of prime people on. Um and that was very well reviewed. Um people like said we should bring that back. We didn't do it this time. Uh but I think for the the summits we can definitely do that. Can I give the tactical feedback on the mobile application if you can add filter on location uh to like filter out booth for example if you're looking for agenda or something like that. \>\> Ah uh \>\> very tactical small thing. \>\> Um if if I'm I assume you don't you're talking about the dedicated mobile app right? Yeah, the the dick mobile. \>\> So, uh if if I have my way, uh that will go away next year and we will just do the web app um as a as its own its own thing. Um so, I mean I I I hope that this was useful to people because I I would love to have this because then I can vibe code whatever filter I want. Um I don't know. Um, yeah, we want to get rid of that vendor. \>\> Yeah, honestly, uh, that app is a slob. Uh, \>\> which one? Well, the the mobile app. It's \>\> Oh, okay. Some people say they liked it because well, for example, it doesn't filter like the past sessions. You'll have to scroll down like I mean \>\> just set the filter and that's it. \>\> Yeah. Yeah. It's rough. Okay. No, this is why I have resisted making an app for like for us even though we can code. I resisted making an app for a long time because they all suck, right? So actually the only app you like is the one that you made. So why don't I just publish all my all my JSON files, right? And then go knock yourself out. So there was nothing stopping you from vi coding your own stuff because the the data was open the whole time. We have MCPS and CLI and um uh people some people did it, some people don't. uh but you could you could always like if if there's any conference where you can vi code your own app it's this this one so why not uh I mean but obviously we should provide a good default and so next year I want to take control of that um because I think I finally understand what we need to show people um and it took me three years to learn this problem because my my sample size is not that high this is only the seventh AIE \>\> the game is really cool the game Um, \>\> people, \>\> a lot of people asked what you spoke about. When do we get your keynote? When do we get you to do an online talk? When do we get to hear your takes? \>\> I did a keynote yesterday. Um, Tuesday. \>\> DJ, \>\> I thought I don't know. I was pretty happy with my keynote. Uh, basically it's like the idea was uh everyone is doing loops. Oo, look at that. Um, everyone's doing loops and um, sorry, I think this is very distracting. \>\> This was just part of uh, you know, the content the CLI is out there. \>\> Yeah. Okay. This is \>\> this for the \>\> So only only Opus decided to do an auto tour thing so I don't have to touch it and it tours through all the talks. Um, okay. So um yeah my talks I don't know I have a podcast I don't need to speak \[laughter\] to talks \>\> um yeah just one other point um I thought of just now was that the breakout workshops are really cool but sometimes 20 minutes is just like not enough to get through anything of sort of substance um and I thought the workshops themselves are really good because you had an actual hour to see how it actually looks in practice as opposed to sort of like one vendor being like Yeah, we have this app and this is what it can do and you just kind of it's very high level. So if they could consider like a mixture of lengths of um \>\> you know we mixed them right we actually did \>\> okay nothing I went to was long. \>\> Yeah. Yeah. So so I mean it wasn't that super obvious but um um \>\> that's on that's on me. \>\> Yeah. No it's it's fine. It's only I know everything right. Um so every day had a workshop uh schedule. So this today this is the the workshop one where people have like this is basically two backtoback sessions this 45 minutes uh and then every day had a workshop room where if you had longer form content I would give you that room so this is workshop this is basically room four had their workshops day um so every day had that but a lot of people just have talks so I did um one of the three or four inspirations of AIE is GDC, Game DevCon. GDC does a lot of one-hour workshops where it's like, okay, here's everything about web slinging in Spider-Man 4 and and they do that and those those are typically what you do in these kind of workshop rooms. Others would be more sort of quick hits like lightning talk like type type things. Um, I would say that attention uh online attention on YouTube is going shorter and shorter in terms of attention spans. So honestly making people condense into 20 minutes is pretty good and we have an entire workshop day 10 tracks of workshop day on the first day as well as three workshop rooms across the remainder of the days. I feel like it's enough. Um, you know, arguably it's can can be up or up or down, but we have something. I Yeah, I do want people to go deep. Many people don't have don't deserve the time, but they want the time, and so I don't give it to them. Hey, um thanks a lot for organizing this. It's been really great for the um for all these subtracks are is all of it going on YouTube. Is it all recorded? Yeah, \>\> every single one. \>\> Yeah. And then maybe this has been asked already but like continuity if there's a track let's say vision and OCR is is there an attempt to make uh there be like a consistent level of depth or like like sometimes there's repeat subject matter between uh subsequent speakers and obviously if if if you're going for you know the YouTube experience maybe having it self-contained in 25 minutes is the best experience but how do you think of like weaving a common thread in like a 4 hour block? Uh I we do actually I I um in previous AIES I would have I would manually do it of like try to sequence and cluster topics together. Uh Jazelle was here. She was in the MCP block of the context engineering track, right? They're all MCP and I I sequence them so they like kind of like se segue into each other because I know the the content of every single talk. Um sometimes it's just impossible if the speaker cannot come in the morning. They have work. Mhm. \>\> What are you going to do? \[laughter\] \>\> So, you try very very hard to have this pristine, beautiful schedule and then people have lives. \>\> Yeah, \>\> it's very annoying. I need them to not have lives. I need them to just come here and only do this. But unfortunately, this just not realistic. Um, so, uh, you know, uh, we we we basically try to optimize, but I think you're right in that YouTube standalone is the format to optimize for. I can craft a narrative with my playlist, which I I set up playlist for every single track that nobody uses. So, I do it, but \>\> Eric for number two. \>\> Yep. Um, so appreciated the new engineer orientation this year. I haven't been the last. You liked it? \>\> Yeah. But I feel like it could have been more thorough. Um, it kind of hit a couple highlights, but like the next day I still wasn't sure if lunch was going to be served or uh like how to manage the schedule and things like that. So, \>\> you know, again, \>\> we can improve this \>\> in the context everything is great, but like \>\> that was one thing that still felt a little bit uh like I could have used some more information before Monday morning hit. \>\> I think again this is related to the previous comment about how next year we will not grow. We will see the same venue, same location, same time, same everything. So that we have a year to like to prepare y \>\> and we now have 7,000 people who have been through this. \>\> Yep. \>\> So uh I think that we just need to optimize our processes. This year my team was just keep like struggling to already organize the actual days. Having an extra load to organize NEO was was a you know was was an undertaking that I think I underestimated. I really wanted more onboarding and in instead it became more networking and it's fine. It's just not not full on boarding. We had an attendee guide that we put together. I don't think most people saw it. It's online. \>\> Okay. Yeah. I didn't see that one. \>\> Yeah. Yeah. \>\> Yeah. again, \>\> we send a lot of emails and not everyone reads the emails. \>\> No, I mean this is a tiny little \>\> No, I I I appreciate that and and I mean uh you know Justin over there is a first- timer and like I I'm sure he could have used better onboarding. I don't know if you went to our NEO the new new engineer onboarding. Yes. No, you did. Yeah. Uh so like uh the NEO came out of this kind of feedback session at the last conference, right? Right. And then we were like, "Okay, well, we went from zero orientation to a little bit and like I think the next one will will get a lot better." But yeah, I I um I didn't know some of the answers that you some of the questions that you had questions for anyway. So, uh it's this process of discovery, I guess. \>\> Yeah. Just one small piece of feedback. So, \>\> thank you. Yeah. \>\> Kevin, I want you to say something Andy. you've you've been to \[snorts\] many of these. \>\> So, um, sorry, I joined late, so it may be that \>\> Q&A FnB. Yeah. \>\> Right. Right. It may be that some of the questions, some of the things that I point out have already been pointed out. So, sorry for that. Um, but it does kind of segue into one of the things that I have is that \>\> I feel as if though there were so many talks and so many interesting sessions that I wanted to be in that at some point I kind of just gave up. I'm like, okay, I can just watch them at home. Y \>\> and like part of it makes sense. Sure. Like I cannot be in two places at once. Um so it was going to happen anyways. \>\> Yeah. \>\> But I do really enjoy the part of talking to the speaker afterwards and being able to ask questions and and and know more about that. And so I feel like I lose that \>\> and it kind of feels that like that sense of okay, I'm already missing two talks, so why not just stay in the expo? \>\> Why not miss all of them, right? Why not miss all of them kind of thing? It's like I'm I'm going to have to sit down and watch them at like 2x when I'm when I'm home. So it's it's okay. So, it's one of those things where it'd be nice to have them a bit more maybe less, right? But more quality or something like that. I don't know. Like the quality was good. That's the that's the thing. Uh it's just yeah, I felt overwhelmed if that makes sense. \>\> So, uh I think what I hear is anxiety, right? FOMO, anxiety, \>\> and it's something I do want to manage. At the same time, it's also just a high class problem. We have an embarrassment of riches. We happen to be in a golden age of AI and we have, you know, too much food at our table. Oh no. Um, you know, the food doesn't spoil. It it lives online on on YouTube. Uh, again, so like um am I is that like that is important to me, but is it the most important thing to me? Probably not because I think you have coping mechanisms um which you you handle just fine. Yeah. um does uh we have other formats of AIE where there is no there's only one track right \>\> that's what the summit is it's only one track so you can attend a track or not attend a track there is no other track and I think that's very simple um that's not what the world fair is fair is the big show \>\> gota \>\> and we want to create the space for everyone to show their the best work of the year and if they want to go to a more dedicated summit they can to they can but this is where everyone shows their best Got it. Got it. \>\> Um, what else should I say about that? Um, and you know, if we were covering finance or real estate or healthcare or something like that's fine. You know, we're covering the fastest moving fields ever. And so, you should be a little bit uh uncomfortable, right? So, if you're completely comfortable, you're probably not being pushed, \>\> right? You're probably falling asleep during the talks. Yeah, \[laughter\] \>\> fair enough. And then I don't know if anyone touched on the app itself. \>\> Uh there were a couple comments. Uh do you like it or hate it? \>\> Um I have mixed feelings. So there's there's \[laughter\] Yeah, I I can't say I'm fully satisfied. Not not going to lie. And obviously I'm going to point out all the bad, but take into account this is within the context of \>\> it mostly worked. \>\> It mostly worked. Exactly. Yeah. Yeah. But things like for example on Android it crashed multiple times. Like it just if you enter and then exit and try to enter it again, it just at least for me it just crashed. Yeah. Yeah. Yeah. \>\> For no reason. Um, if you enter a talk to see the details of that talk and you go back, you get a loading screen for like several seconds. \>\> Yeah. \>\> What's it jump to the top, \>\> right? Oh, that I didn't didn't jump to the top for me. Holy moly, that's even worse. But I don't know what's loading \>\> for for you. It it jumps to the you lose your position and scroll. Okay. \>\> Right. Um, you sometimes get the loading screen as well when you're just scrolling down, which I assume is just like fetching more talks from the server, but it's like, can you not pre-do that while I'm scrolling, right? Like it feels it feels very 2015. \>\> So, we we we Yeah. Okay. Uh, but keep going. \>\> Okay. Yeah. Yeah. So, um, the my schedule um screen, I would expect it to be very similar to the regular like schedule, just like a filtered down version of the event, but I can still search, I can still filter. I can't do that. Uh not only that, the my schedule tab you scroll from the top like from the very first day even though all those other talks have already ended you still have to scroll through all of them to get to today whereas that doesn't happen on the regular like all sessions kind of kind of tab. \>\> Okay. \>\> Yeah. Do you want me to keep going like there's a \>\> Yes, I do. Uh because all this is going straight to my uh my team so that we can we can fix it for next time. \>\> Awesome. Awesome. um uh you can't save currently ongoing sessions to your schedule. So if there are multiple things going on live and that's happened to me. I'm like okay the let me look at what's next and it just so happened to be one minute past you know the the time and so I couldn't save them. So it's like okay I want to go to these two let me save one and go to the other and I couldn't do that. Um then there's also no syncing between the web and the app. I can't log in into the web. So if I went on the \>\> because they're different code bases. Yeah. \>\> Right. Right. So, it would be super nice to have that so I can no matter what UI I I use, I see everything synced across. Yeah, \>\> that's just nice to have. \>\> Um, for the booths, it's it just says booth \>\> like doesn't have a name. \>\> Yeah, it just says the title of whatever they're doing, but then on the calendar icon and it doesn't have the address booth booth \>\> and it's like okay. And so some many of them on the message like when you actually click, they've gone to the lengths of okay, we are at booth whatever blah blah blah. This is our company. but some haven't. So, it's like, "Oh, we're doing this thing." And it's like, "But where?" You just don't know. There's no way to know. Um, so yeah, that's And then there were I I noticed there's a chat and a Q&A features within the the details. That's super nice, but I don't think anyone's ever Yeah. \>\> used any of them. So, maybe something you can just remove. \>\> Um, then the filters themselves, not too cool because when you open the filters, it just I think it just loads like a bunch of the companies. And so for you to actually \>\> it's unoptimized. Yeah. \>\> Yeah. \>\> Yeah. \>\> Um and then maybe a nice to have would be potentially like a matching system. But this is very nice to have something where it's like I'm interested in seeing people \>\> who are similar to me and I want to talk to them potentially or I want to meet more founders or I want to maybe meet people from like big tech that kind of thing where it's like he's the recommended people you could talk to. the leadership track had a a prototype basically of our matching system that we set up. Uh and we basically built a Facebook where we sort of deep research every leadership attendee uh and and sort of sent that out. Um we I want to invest more in that direction because I've seen it work in other conferences. I've never seen the chat work. Uh so yeah, I agree with that. Long story short, that is a v that is an external vendor that we that we purchase. Uh and next year I want to rip it out and do our own. \>\> Got it. \>\> Right. So, uh, to the extent that you've seen, uh, the website, this one is this one is controlled by me. Um, and we have the we have the sort of canvas like thing and we have the performance. \>\> The website's super nice. Like it's much easier to navigate. \>\> Yay, this is me. \[laughter\] So, so I I I'm somewhat trying to build confidence as an organizer of that we that we not just organize the events, but we also do the event technology which are completely separate uh disciplines effectively. It's not like I'm I am an engineer. It is just that this is a design thing that I wanted to completely uh collect the specs. So everything that you just reported, we were recording this session. Those become the the sort of requirements document for me to do next year's thing. \>\> Fantastic. Fantastic. I know I threw a whole bunch of things at you. \>\> Yeah. Yeah. I'm glad that you can probably find Yeah, you can probably find my email if you have more that you'd think about later. Uh Swix at AI engineer. Uh yeah, we want to hear all this and and build a like you know engineers deserve a good app. Uh that said uh like I said before to other people. Um we have always open sourced our data. So if you wanted to vibe code your own you could you could have done that \>\> and and that's a great thing by the way like I wish more like events did that cuz it's like \>\> because why should you like Yeah. I always say like I have never liked a conference app that I that anyone else made. I just like the one that I make and I I assume it's the same for every attendee. Just to extend on that last idea, you can gamify it and put it like the marketplace where people can upload their application. \>\> This is it. This is everybody that submitted. There's only three uh four. That's it. There's not that many submissions. \>\> There's opportunity to grow. \[laughter\] Hey Sean, thanks again for doing this. As you know, um I work for a nonprofit in the sports industry, so um we don't have a lot of money, but we do choose to invest pretty heavily in getting here every year. I've done it for four years now. Um \>\> thanks for coming back and bringing your team. \>\> Yeah. So I maybe I'll start with that. Um there are is at least one organization that treats this as part of our annual clock tick. This is the investment that we make in our team. We spend tens of thousands of dollars to bring our people here because of you. So for all the feedback and all this kind of thing that you're going to get that's constructive like thanks for making it all happen. It goes beyond just the the content and the things that are here and the speakers. This is actually how we invest in our team. So thanks for being uh sort of a pivotal part in that. \>\> I have a related so persuant to the word invest. I have a related suggestion but I want to I want to hear your feedback first. Um and just the way that we approach that is like when when we come here, we actually um also get together a month before to prepare for being here. \>\> Yes. That's what I want in every team. \>\> Yep. So we've created our own little internal level up summit to make sure that everyone is here to get the most out of the time that we're together. Um and then when we're here, it is like we rent a suite, we get beer and pizza every night, and we basically hack on the things that we heard and learned during the day. So just recognize that it's different than what you see inside these four walls. Uh there's a lot of other stuff that at least one organization is doing with it. So thanks for being part of it. Um I just want to applaud the part where you were u the leadership track is significantly better this year really \>\> has been previously. Um like content fantastic I think. Um \>\> voice opinion of one. Can you can you call out uh sessions that you felt like made it? \>\> Um not in particular, but they um in previous years it has felt more salesy or um hungry for relevance and uh we have people who are just legitimately doing awesome things who are able to get up and talk and speak credibly about it. Um I I was just very impressed by that and it's actually it is more than opinion of one. I heard that from several other people as well. \>\> Can I just say \[clears throat\] um I had nothing to do with that. I cannot take credit for any of it. It is simply that we are we have been around longer therefore people have heard of us therefore they come forward. It's all inbound. We didn't really I mean I solicited some of them \>\> but most of it it is people stepping up saying I have something to share. \>\> Yeah. \>\> Because we've been around enough that like we're established now. \>\> Yeah. Well something must be working right for that to happen. Um in the I think the very first year there was some kind of birds of a feather concept where you would sort of set \>\> Yeah. table topics. Yeah. Table topics. That's right. And it sound kind of similar to what you were talking about maybe doing in the future as part of the leadership track or a second content uh expo area. \>\> Oo. Uh \[clears throat\] maybe \>\> the idea of going and participating. \>\> Yeah. \>\> And not rather than just listening to participate \>\> time place on a topic uh is \>\> I mean it's like what it is is unconference, right? Have you \>\> have you been to one? Do you like it? \>\> Um it's been a decade or something but yeah. So \[clears throat\] people want us to do an unconference. The problem is I worry that we don't sell tickets. If \>\> you come and you don't have any speakers but yourselves. \>\> Yeah. I I wouldn't go if that was it. I'm just saying that like there's room in there. Uh I would find that compelling at some point to wrap as a a thing a niche. \>\> Yes. Uh yeah. \>\> May maybe it's different from what you were envisioning with the cont. look like um okay especially for the leaders we do actually want maybe next year maybe the year after I don't know to do a dedicated leadership retreat uh like only leadership go away to like a some Napa or something and sit down for a weekend uh to me there I can do the unconference there I can do the off off the record chats I think there will be a market for that um yeah I mean but at these things content is king we want to get as much good work online as possible um well produced with with these guys and uh for you to meet the speakers in person. \>\> Yeah. Uh the the last thing I was going to share is um and you don't really have to react to this, but just um over the years of you know for a long time we've had things like South by Southwest that started out scrappy and with a certain vibe and eventually the the ratio of \>\> uh sport coats to hoodies started to take over and that will naturally happen in in every place that becomes successful like this. Um a little bit of me is going to be sad when that happens. So I I really appreciate the heart and soul that you put into all this. Do you don't you notice it's gotten more uh chill? I I So, okay though, uh my index for this \>\> I don't think it's happened yet. \>\> My index for this is number of people wearing costumes, \>\> right? We had two orcs. \>\> Yeah. \>\> Walking around. \>\> Okay. We had a guy in a T-Rex suit. \>\> We had the mother duck ducks outside. I don't know if you saw saw the ducks. Uh we had a we had a shark. Um, so the the the the costume index, there's been three Worlds Fair. The costume index was two in the first World's Fair, zero in the last year, and then now it's like nine. Um, so I think we're going in the right direction. We had I mean, okay, then we get a bunch of free costumes with the the flash mob. Did you see the flash mob? \>\> I didn't. Oh my god. Uh, yeah. Anyway, I want to be I told uh Eve, we had Eve, the head of design for Y Combinator, speak at the design track. I was like, I want to be the conference where you would do things that would be unprofessional in the other ones, but it fits here because it has the community vibe because you can be yourself. \>\> That's fantastic. \>\> Uh so yeah, I mean to some extent like you know I will not run AIE forever. Uh but I think if we set the community expectation um that uh you know that that this is what we do. It's like Comic-Con. Comic-Con has like you know there still you still have people it hasn't been taken over by the suits and I think that is important. We do still want to serve the suits. I think there's a there's a place for them. Uh but there's you know a floor of the Mosone separating the suits from the here. \[laughter\] I think that's okay. Yeah. Um what else can I say about that? Vibe and culture. I think um one of the most important things in our CFP application uh the application to speak is we have all these tracks that we are interested in but the last thing in the track is my talk is weird and doesn't fit and I always want to be have a place for weird stuff and so um Eric Meyer's talk from last night came from came from that John Auster's stuff came from that sometimes you just have a you really want to just share and it doesn't fit yet, but if it's cool, uh we'll try to give it a platform and a place. Uh kind of on that note, I had a question about uh speakers. So that's why I'm here. I think that the speakers, the quality of speakers you are able to assemble here is amazing. Um I I applied my whole team as speakers and so like five applications total and I was I was accepted for the remote track and I guess in short I'm I'm a little bit almost like buttth hurt that my teammates who we work in spatial reasoning and have a a a product for home builders that just raised series C where we have uh paying customers using novel computer vision and uh spatial intelligence tools that are um the first time that's been done in architecture. And I wanted to ask like in in the applications like these guys who I I submitted their applications. They're not Twitter famous. Um they're not from Silicon Valley. They're architects turned PhD machine learning engineers. How can I better like tell their story in the application to get a better chance of them getting picked up to speak on that stuff? \>\> Uh uh so again I'm sorry for not uh that we can't select everybody. Um you were weightlisted or you were selected for online. Uh let's for online. Yeah. \>\> Yeah. Uh so uh you know we we had two 2200 applications for effectively 300 spots. \>\> Um and all of those 300 spots we had to we also invited people. It's not just all uh people that come in through the CFP. So basically it's very very competitive. We can't accept everybody that like does meaningful work because there there's just there's too many. Um I will say I think um being somewhat frontier being uh uh being unique in the application um that happens to tie it to a broader trend that we can tell um does help it more. I don't know if um I don't know if that directly applies to your situation. Yeah. Because basically there there's a lot of like early stage companies that have re revenue that h they have innovative things and we only have 10 talks in the vision track. \>\> Yeah. \>\> Right. And some of that immediately goes to people with proprietary models like the perceptron talk uh that we have there. Uh and some of them goes go to leading companies that are somewhat Twitter famous because that puts that puts butts in seats. \>\> Yeah. \>\> Then we have some leftover for \>\> well people who like let's let's see how they do. \>\> Uhhuh. So, the better you can guarantee the review committee, I'm on every committee, but then uh I work with uh reviewers that that we trust, usually past speakers, um to to to vet and and and vote on on who should be in. Um if you can guarantee that this will be a banger basically. \>\> Yeah. Yeah. Yeah. \>\> Uh then that would be great. Uh that's why we often that's why that's why um applications are not blind. We do encourage you to submit prior talks. \>\> Yeah. and we do watch the talks \>\> and if we go this guy is really goddamn good, we'll fight for you. \>\> That's what I was going to ask too. So it like let's say for next year that's a good goal to try and like just \>\> have have talks recorded at other meetups, other conferences \>\> and like some proof of like even if it it doesn't have a thousand views doesn't matter. Yeah. \>\> But like I watch you like you're com you're you're you know how to do a good conference talk. We have that quality filter here. It's a it's a very high bar to be here. Yes. It's a lossy bar. Like some people get through that should not get through. I know. \>\> Um but mostly we're known for having a very high bar. So try to just remove all doubt that you meet that bar \>\> and you'll be in. \>\> Awesome. Thanks a lot. Anybody else? Um so uh first of all I want to just thank you because I am attending because I um submitted a demo to the ACM agents project uh agents and um agentic systems conference and uh we're able to um and I I I wanted to just comment that I think it's great that you were able to um provide access to this conference for students and for academics and people who are kind of doing that. Um \>\> thank you. The more you can do that, I think the better because a lot of the people who were in that when I was attending that meeting, a lot of the people who were there were just these young, brighteyed, really excited people who are just starting their careers. They're doing some really interesting stuff from a conceptual point of view. And I I think that um having a little bit more of that in the community may also be just a good you know um catalyst to get things moving to kind of get things get \>\> Yeah. Um thank you for saying that. Uh I was this is the first year that we've done the poster sessions um so AIE was started kind of in opposition to new CML clear uh because we basically wanted to do everything that they don't do. for example, don't cut off submissions six months before the the conference date because that guarantees everything's out of date. Um, but we want to bring academ academic academia to industry uh in the same way that Nurus brings industry to academia and I think uh there is a lot of synergy there. Kais was just the easiest to work with by far. Uh so we have this partnership. It's very it's been very successful. We've been invited back to Kais and Kais has been invited back to us. So we keep going. Um I think the the just the question is like well okay this one we just use KAS as our selection process right we didn't do any selection whatsoever we're just like we just outsource it but at some point we're going to have a peer review process um I've been introduced to the ACM presidents uh do we start publishing our own journal to give people a publication spot I think to some extent um I like that we are not accredited because that means that nobody breaks any exclusivity rules by presenting here because they can just double present event. If they're already in the Bay Area, that's good. Uh if they just want to double dip on their work, uh and present in front of an industry audience, that's good. Uh and yes, I like the the energy of the students. Uh because a lot of people actually want to hire them. Uh so, uh Worlds Fair is basically the only place I can do this because I have the real estate. Uh and so we'll we'll try to do it more, but I I think it was relatively successful this time. \>\> Absolutely. Thank you. Uh that might be it. We have the AIE Singapore team that just came in. Uh they also did the networking app. Some of you asked about the networking app. It's him. Uh it's it's Sharina Grim. Uh otherwise I think that is it. I'm going to call it and I have to go and do the startup battlefield. But thank you so much for coming. Come back next year. Hey