Apr 24, 2023 · 40m · a16z
Beyond Avatars: How AI is Reshaping Online Identity (Danny Postma and Sinead Bovell)
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
Host Steph Smith leads a discussion with futurist Sinead Bovell and indie startup builder Danny Postma on how generative AI is revolutionizing digital identity, commercial fashion modeling, and software entrepreneurship. Together, they analyze the economic shifts, technical architectures, and future retail applications of synthetic media across consumer and business domains.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The host holds 44.7% of the talking time here. How this is scored →
The host as informed peer 3.7 Guest teaching 4.8 Guest disagreement 0.4 The host pushing back 0.6
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 38:11 Dismissing startup feature idea
Danny directly dismisses the host's startup idea by asserting with 100% certainty that Canva will capture that capability within three months.
Hardest push from the host ▶ 12:36 Reframing photo realismSteph pushes back on the narrative that AI photography removes truth from fashion, highlighting how pre-AI commercial photos are already deeply fabricated through heavy editing.
Biggest teaching moment ▶ 34:15 Demystifying AI app architectureDanny educates the host and listeners by revealing that quality production tools require complex multi-model pipelines rather than simple API wrappers.
The host holds their own ▶ 37:34 Formulating brand-constrained prompt modelSteph demonstrates product expertise by identifying a specific unfulfilled niche for self-serve, brand-constrained generative assets.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| Legal Disclaimer and Podcast Bumper | 3 | 4 | 0 | 0 | Steph introduces the episode context comparing CGI influencers like Lil Miquela to new generative AI tools. Sinead explains her 2020 Vogue article on deepfakes and how Generative Adversarial Networks pose an existential shift for fashion e-commerce. | |
| Danny Postma's Viral Deep Agency Experiment | 3 | 5 | 0 | 0 | Steph asks Danny about the viral reaction to his Deep Agency launch. Danny educates the host on the technical intersection of DreamBooth and StyleGAN for building non-existent models and fitting virtual garments. | |
| Industry Adoption and Authenticity vs Realism | 5 | 4 | 1 | 2 | Steph offers an insightful counter-perspective on traditional media authenticity, noting that billboard photography is already heavily altered. Sinead agrees and elaborates on how cultural demands for raw photos collide with AI adoption. | |
| Economics of AI Modeling and Wage Pressures | 3 | 6 | 0 | 0 | Danny outlines how 60-cent model training democratizes content creation for small businesses. Sinead provides an economics lesson on technological deflation, wage pressure, and ethical concerns over licensing human likeness. | |
| Danny Postma's AI Product Suite and Scale | 4 | 4 | 1 | 1 | Steph lists Danny's viral product portfolio and cites live statistics. Danny playfully corrects her user count figure from 450k to 1M and explains how remote teams use Headshot Pro to standardize team headshots globally. | |
| Monetization Lessons and Creative Frontiers | 3 | 5 | 0 | 0 | Steph inquires about user willingness to pay and market dynamics. Danny shares how Lensa's price drops disrupted the market and describes the rapid pace of open-source AI updates. | |
| Future of Digital Identity and Virtual Try-Ons | 3 | 5 | 0 | 0 | Sinead and Danny describe future creative paradigms, such as AI serving as pre-production image synthesis and interactive social media carousels featuring virtual twins. | |
| Strategies for Keeping Up in Fast-Moving AI | 4 | 4 | 0 | 1 | Steph asks both guests how they stay updated on research and queries whether solo products can maintain moats against larger competitors. | |
| Building Technical Moats and Distribution Channels | 4 | 6 | 1 | 1 | Danny delivers practical advice on distribution beating pure development. He explains his technical moat strategy, revealing that Headshot Pro chains 15 distinct AI models together. | |
| Startup Advice and Niche Utility Tools | 5 | 5 | 1 | 1 | Danny warns builders against competing with incumbents like Adobe or OpenAI. Steph pitches a brand-constrained image generation tool idea, which Danny predicts big tech platforms like Canva will release shortly. |
Statements from this episode (18)
Bovell: E-commerce modeling is vulnerable to AI due to low pose variability
“And the clear cut for me was fashion especially when it comes to e-commerce a line or a segment of fashion where it's not necessarily about these creative big movements for a fashion model, but instead just small variations in the posing. And that in some ways…”
Postma: Built Deep Agency prototype in 24 hours after viral tweet
“My product wasn't ready yet. Deep agency, the tweet went absolutely viral. So within 24 hours, I had to come up with something to put live because otherwise there wasn't a website online.”
Postma: AI fashion modeling needs industry domain expertise, not better models
“So the technology is there at the moment, I believe it's up to a company that understands the fashion industry to take it to the other level.”
Bovell: Zalando demonstrated AI model proof of concept in 2019
“So if you look at e-commerce and a company, although we might not be as familiar with Zulando in America, they're quite a big fashion giant in Europe. For them to be researching this technology and showing very viable proof of concept as early as 2019, we know…”
Bovell: H&M employs thousands of data scientists using AI
“And we can also say a company like H&M, they have Thousands of data scientists that use AI to do things like forecast trends and understand and analyze supply chain.”
Postma: Fine-tuning an AI face model costs 60 cents
“So currently to deep train a model on someone's face to be able to generate photos After that, it costs 60 cents to generate this. Yeah. So not even a dollar. And then I think per photo you're at a fraction of a cent right now. So you could generate hundreds o…”
Postma: Small clothing brands are primary users of AI models
“So I believe this is from what I've seen talking to customers, people, the number one question in my life chat is from mom and pop shops with close clothing brands who do not have the funds to hire models, to put on clothing for the web shops, who I think prob…”
Bovell: AI will depress wages in modeling and programming
“We're going to see this across the board with AI. It's going to put downward pressure on a lot of different wages from modeling to programming, but that's where it becomes a little bit challenging.”
Postma: Lensa generated $40 million from AI profile pictures
“I believe Lenza did forty million dollars with that.”
Postma: HeadshotPro has generated 1 million AI corporate headshots
“I think it's a million now, because I have to manually update it the next two weeks live now.”
Postma: Lensa destroyed the AI profile picture market
“So yeah, Lensa basically killed the whole market after they went live.”
Bovell: E-commerce campaigns will let shoppers drop virtual twins into photoshoots
“Where we each get to be the model in the shoots that we see, so instead of just looking at the campaign, there might be an interactive option where you could drop in your own virtual twin to see how you would personally look in that image.”
Postma: Social media will integrate AI clothing try-ons within years
“And I wouldn't be surprised if any social media is going to do this soon, because Instagram already has photos of your face, right? They could already train a model on your face in that kind of sense. And you could put the clothing on it that they put in their…”
Postma: Uses ChatGPT to summarize academic machine learning literature
“Also for me, I'm dabbling in all the white papers. I don't have an academic background, so I'm using ChatGPT a lot to summarize those things and understanding it mostly.”
Postma: Lensa matched my startup's total lifetime revenue in one hour
“Seeing Lenza swoop in one month after we actually launched this already and just did our revenue in one hour or something.”
Postma: HeadshotPro stacks 15 distinct AI models to build a product moat
“What I'm doing with HeadShield is basically it's 15 different AI models stacked on top of each other in like a sequence of doing things together.”