Brian Chesky on How AI Will Reinvent Airbnb — And Why Most Companies Are Using AI Wrong
Airbnb's CEO argues that most companies are bolting AI onto existing products when they should be redesigning products from scratch around AI — and he's rebuilding Airbnb to prove it.
Top Claims — Verdict Check
Most companies are using AI wrong — they add chatbots and features to existing products instead of reimagining the product around AI
🟢 Real“Adding a chatbot to your existing product is like putting an engine on a horse-drawn carriage. You need to build the car. Most companies are bolting AI onto the old design. [representative paraphrase]”
AI will turn Airbnb from a search engine into a concierge that knows you and plans for you
🟡 Partially True“The future of travel is not searching through 7 million listings. It's telling an AI what you want — "a quiet week with my family near the coast" — and having it design the entire trip. [representative paraphrase]”
The best product design comes from designing for one perfect user rather than statistical averages
🟢 Real“We don't design for personas or averages. We design for one specific person and make it perfect for them. If it's perfect for one person, it'll be great for many. [representative paraphrase]”
Airbnb runs like a startup despite being a $80B+ public company — small teams, founder-led product decisions, no middle management bloat
🟡 Partially True“I'm in the details of every major product decision. We have a functional organisation, not business units. I review every design. That's how you maintain quality at scale. [representative paraphrase]”
AI will make small teams dramatically more productive — the era of 1,000-person engineering orgs building mediocre products is ending
🟡 Partially True“A team of 10 engineers with AI tools will outperform a team of 100 without them. The bloat era of tech is over. [representative paraphrase]”
What's Real
The 'bolt-on AI' critique is the most important product insight in this conversation and it's directionally correct. Most enterprise AI deployments in 2024-2025 were chatbot layers on existing products — Microsoft Copilot, Salesforce Einstein, Adobe Firefly — that add capability without redesigning workflows. The companies seeing the largest AI-driven revenue growth (Cursor, Perplexity, Midjourney) are AI-native products designed around the technology, not incumbents adding features. Chesky's design philosophy has results: Airbnb's 2023 product redesign (Categories, flexible dates, AirCover) drove the company's strongest growth quarter in two years with fewer engineers than Booking.com's product team. The founder-led product approach — Chesky personally reviews major designs — produced Airbnb's visual refresh that received widespread design industry praise. The small-team productivity thesis has evidence: WhatsApp served 450 million users with 55 engineers before the Meta acquisition. Instagram had 13 employees at 30 million users. AI amplifies this dynamic.
What's Hype
The 'AI concierge that plans your entire trip' vision sounds compelling but has deep execution challenges Chesky doesn't address. Travel planning requires integrating real-time availability, pricing, ground truth about locations, and personal taste calibration that current AI systems handle inconsistently. Airbnb's own AI search improvements — while better than keyword matching — still rely on users scrolling through listings. The gap between 'tell the AI what you want' and 'AI books your perfect trip autonomously' requires solving reliability, payment integration, and liability for bad recommendations. The 10-person-team-beats-100 claim is true for some product categories and misleading for others: Airbnb's Trust & Safety team alone requires hundreds of people handling disputes, fraud, and guest/host conflicts that AI can assist with but can't resolve. The 'no middle management' framing obscures that Airbnb had significant layoffs in 2020 (1,900 people, 25% of the company) before arriving at this 'lean' structure — it's not a design choice that scales painlessly.
What They Missed
The platform dynamics that make Airbnb's AI vision harder than it sounds. Airbnb has two customers — hosts and guests — with often conflicting interests. An AI concierge that optimises for guest experience (lowest price, best location) may conflict with host interests (highest price, minimum effort). This two-sided marketplace tension doesn't disappear with AI; it intensifies because AI makes the optimisation explicit. The regulatory environment for AI-powered booking is also absent: EU regulators have already scrutinised algorithmic pricing in short-term rentals, and an AI that autonomously books trips raises questions about price discrimination, hidden fees, and consumer protection. The Southeast Asian short-term rental market specifically has different dynamics — in Malaysia, Airbnb competes with Agoda, Booking.com, and local platforms like OYO that operate differently. Chesky's product vision is designed for the US market where Airbnb has dominant brand recognition; the AI concierge approach requires trust that Airbnb hasn't fully earned in APAC.
The One Thing
The real AI product opportunity isn't adding AI features to your existing product — it's asking what your product would look like if you designed it from scratch assuming AI existed.
So What?
- Do the 'blank slate' exercise for your own product: if you started today with AI as a given, would you build the same thing? If not, you're carrying design debt that competitors without legacy won't have
- Chesky's 'design for one person' principle works with AI: instead of building features for statistical averages, use AI to personalise the experience for each user. This is tractable now in ways it wasn't pre-AI.
- The small-team advantage is real for Malaysian startups: a 5-person team using Cursor, Claude, and v0 can ship product at a pace that used to require 20-30 people. Capital efficiency is your structural advantage against well-funded competitors.
Action Items
- 1Run a 'blank slate' workshop with your product team: spend 2 hours designing your product from scratch assuming AI is the primary interface. Don't constrain yourself to current architecture. Compare the result to what you have. The delta is your AI product roadmap.
- 2Identify one user flow in your product that requires 5+ clicks to complete. Redesign it as a single natural-language input that AI resolves. Prototype it in a week — not to ship, but to test whether users prefer it.
- 3Audit your team structure against Chesky's principle: how many layers exist between a product decision and execution? Each layer is friction that AI-native competitors don't have. Remove one this quarter.
Tools Mentioned
Cursor
AI-native code editor — exemplifies the "design for AI" principle Chesky advocates rather than bolt-on AI coding assistants
Figma
Design tool Chesky references in his product review process — his design-led approach is central to Airbnb's product quality
Workflow Idea
Do a quarterly 'Year Zero' exercise. Pick one feature or product in your portfolio. Spend two hours designing it from scratch as if you were a new startup with access to current AI capabilities. Don't think about migration, legacy code, or existing users — just design the best possible version. Then compare it to what you have. The gap tells you where your design debt is accumulating. Chesky does this for Airbnb's core flows; you should do it for yours.
Context & Connections
Agrees With
- tobi-lutke
- dylan-field
Further Reading
- Brian Chesky's 2023 Airbnb shareholder letter — his clearest articulation of the founder-led product philosophy
- The Masters of Scale podcast episode with Brian Chesky — Reid Hoffman interview on Airbnb's design culture