CHAPTERS
How the Consumer AI Top 100 is built (data sources, web vs. mobile, usage not revenue)
Justine and Olivia explain the purpose of the Consumer AI Top 100 and how it’s compiled to reflect what people actually use. They break down the methodology: ranking AI-native products across the world using web visits and mobile MAUs, with an emphasis on total usage (including free) rather than revenue.
- •List is published every six months; this is the 5th edition
- •Sample frame spans essentially all websites and mobile apps globally
- •Web ranking = global monthly visits (Similarweb); Mobile ranking = monthly active users (SensorTower)
- •Top 50 web + top 50 mobile, limited to AI-native companies
- •Usage is the signal (not paid conversions), aiming to capture consumer attention/imagination
Who’s new this cycle—and why the ecosystem is stabilizing
They discuss changes in new entrants, noting that web is the best lens for longitudinal shifts while mobile is more volatile due to app-store policy changes. The number of newcomers is shrinking, suggesting the consumer AI landscape is settling into more durable categories and leaders.
- •Mobile list volatility partly driven by App Store policy enforcement
- •Earlier mobile lists were dominated by ChatGPT copycats; crackdowns reshuffled the list
- •Web list had 11 new names (down from 17 last edition)
- •Fewer newcomers suggests stabilization and more consistent winners
- •Emerging clusters still create openings for new breakout products
Companionship still dominates consumer AI usage
Companion/roleplay apps remain a major driver of consumer engagement, with multiple new entrants joining established leaders. The hosts highlight how large and persistent this category is relative to others in consumer AI.
- •Three new companionship names: Juicy Chat, Joy, R Dream
- •Established names include Character.ai, Janitor, Spicy Chat, Polybuzz, Crushon, Adot, Candy AI
- •Companionship occupies a notable portion of the Top 100
- •Category continues to be hard to displace (especially on mobile)
- •Signals sustained consumer demand beyond “general chat” assistants
Creative tools as a core pillar (and the enduring ‘killer apps’)
They frame creative tooling—image, video, audio, and related workflows—as one of the long-standing consumer AI pillars alongside general assistants and companionship. Even as new categories emerge, creative tools remain prominent due to clear user value and shareable outputs.
- •Creative tools historically one of the most reliable categories on the list
- •Many top properties focus on content creation workflows, not just raw models
- •Creative use cases can tolerate more novelty/experimentation than accuracy-critical domains
- •Category remains competitive with frequent iteration and feature launches
- •Sets context for later discussion on prosumer/enterprise spillover
Big Tech’s presence: Google’s breakout across multiple properties
For the first time, Google appears meaningfully in the rankings due to changes enabling separate domain-level measurement. They detail how multiple Google properties—consumer and developer-facing—earned top placements, driven by strong distribution and product momentum.
- •Domain structure changes allowed Google properties to be independently ranked
- •Gemini ranks #2 on web behind ChatGPT; ~10% of ChatGPT’s web traffic
- •Gemini is much closer on mobile (about half of ChatGPT), driven largely by Android
- •AI Studio (developer sandbox) lands in the top 10
- •NotebookLM reaches #13 and sustains traffic long after its viral moment
Google Labs as a ‘consumer sandbox’ and Veo 3’s traffic spike
They discuss Google Labs as a multi-product experimental hub and speculate that Veo 3 largely drove recent traffic growth. The chapter highlights how breakout model launches can lift an umbrella destination that hosts multiple experiences.
- •Google Labs ranks on the list and bundles several experimental products
- •Includes Veo 3 video model; also Doppel, Portrait, Whisk, Project Mariner
- •Traffic spiked ~15% in the month Veo 3 launched
- •Contrast: AI Studio is developer-facing; Labs is consumer-facing exploration
- •Launch-driven demand can meaningfully reshape rankings within a single cycle
Chinese AI companies: three patterns of global presence
Olivia outlines how China shows up on the list in distinct ways: domestic-only products, China-built products aimed at global users, and products that succeed both inside and outside China. Regulation and access constraints shape competition and adoption dynamics.
- •Domestic China ecosystem exists partly because ChatGPT/Claude are restricted in China
- •China-focused assistants: Quark (Alibaba), Doubao (ByteDance), Kimi (Moonshot) rank highly (top 20 on web)
- •Population scale + limited Western competition boosts usage
- •Global-facing China-built products: DeepSeek; video/image tools like Hailuo, Kling, SeaArt
- •Hybrid adoption example: Manus with traffic across Brazil, US, and China
‘Vibecoding’ breaks into the Top 100—and shows unusually strong retention
They cover the rapid rise of vibecoding platforms and what the list reveals about their adoption. Beyond traffic, they discuss revenue retention data suggesting many users increase spend after initial adoption, indicating real utility and potential enterprise/prosumer pull.
- •Shift from Bolt being the only representative to Lovable + Replit joining the main list
- •Lovable reported ~$100M ARR; discussion focuses on whether growth is durable
- •Consumer Edge cohort analysis shows ~100%+ revenue retention in first 3 months for leaders
- •Retention strength is atypical for consumer/prosumer—may hint at team/enterprise usage
- •Vibecoding hasn’t meaningfully appeared on mobile yet, but could in future editions
Builders vs. what they build: the hosting-domain traffic puzzle
They explain that for vibecoding platforms you can observe both the maker traffic and the traffic to hosted outputs, and maker traffic is larger. They propose two interpretations: serious projects move to custom domains, or many projects are personal/internal tools with low public traffic but high private value.
- •Platform domains enable separating traffic to builders vs. hosted apps/sites
- •Maker traffic > traffic to outputs for both Lovable and Replit
- •Hypothesis 1: high-traffic projects migrate to custom domains (traffic disappears from platform domain)
- •Hypothesis 2: many outputs are personal software with limited public visits
- •Implication: usage value may be undercounted by public traffic measures alone
AI All-Stars: the products that never left the Top 100
Olivia introduces “AI All-Stars,” products that have appeared on every edition of the web list. The segment highlights which categories have durable consumer demand and what that says about defensibility beyond having a proprietary model.
- •14 companies made every one of the five web lists
- •General assistants: ChatGPT, Perplexity, Poe; Companionship: Character.ai
- •Creative tools: Midjourney, PhotoRoom, Leonardo, Cutout Pro, Veed, ElevenLabs
- •Productivity: QuillBot, Gamma; Model hosting: Hugging Face, CivitAI
- •Durability suggests strong distribution, workflows, and community—not just model advantage
Network effects and why UI/workflows matter as much as models
They argue that in consumer AI, product experience and community can be decisive because models are increasingly accessible via APIs or open source. They explore network effects beyond “more users → better model,” including libraries, marketplaces, and team-based lock-in.
- •Many all-stars are model hosts/aggregators rather than proprietary model builders
- •Non-data network effects: communities around models/LoRAs/datasets (Hugging Face, CivitAI)
- •Marketplace/library effects: ElevenLabs voice library grows value with more contributors
- •Workflow depth (templating, pipelines, creation flows) increases switching costs
- •Even model-centric products can benefit from classic data feedback loops
Prosumer-to-enterprise migration: bottoms-up adoption and team expansion
They describe how several consumer AI tools are ‘graduating’ into enterprise settings via team plans and organizational sharing. Bottoms-up adoption—individuals trying a tool and then bringing it into work—creates an easier path to enterprise revenue than traditional top-down sales.
- •Examples of consumer tools expanding into teams: Gamma, ElevenLabs, PhotoRoom
- •Team templates/design libraries increase stickiness and reduce churn
- •Individuals act as self-serve champions inside organizations
- •Freemium + consumer adoption can convert into enterprise contracts once usage spreads
- •Prosumer growth can also feed back into more consumer adoption via social diffusion
Biggest takeaways + what to watch next (Grok debut, verticalization, new categories)
In closing, they reflect on how the list evolved from early chaos to a more normalized leaderboard with recurring winners and fewer newcomers. They predict continued ‘verticalization’ across general assistants, highlight Grok’s strong debut, and point to likely future breakout categories driven by improved model reliability.
- •Early lists were highly volatile; newer editions show stabilization and recurring leaders
- •Users increasingly pick different assistants for different tasks (continued verticalization)
- •Grok debuts strongly (#4 web) and drives notable web and mobile presence; Meta AI also dents web
- •Expect growth in accuracy-critical prosumer tools (spreadsheets, slides, financial models, email) as reliability improves
- •Potential future breakouts: consumer health, edtech, personal finance, AI-native social, structured coaching/therapy/wellness
Closing: check the full list and share products that surprised you
They encourage listeners to explore the complete report and discuss their own favorite AI apps. The episode ends with an invitation to comment on omissions and to return for the next list in six months.
- •Call to action to read the full Consumer AI Top 100 report
- •Prompt: share favorite products that made the list
- •Prompt: mention products you expected to see but didn’t
- •Reinforces the six-month cadence for future editions
- •Positions the list as an ongoing snapshot of real consumer behavior
