July 2, 2026 · One Developer vs. the Consumer Internet
Read the series manifestoTarget 01: Rebuilding Airbnb as a Single Developer
Why Airbnb is the first target in HAAM's One Developer vs. the Consumer Internet series, what the independent rebuild will include, and where the real platform remains impossible to fake.
HAAM Rebuild
STAY/01
Search. Compare. Book. Travel. Host. One functional loop, built independently by one developer with AI.
Where
Forest cabin
★ 4.92€126 night · total shown
Old town loft
★ 4.88€148 night · total shown
Sea wall studio
★ 4.95€94 night · total shown
Courtyard room
★ 4.90€112 night · total shown
Compare 3 stays
Evidence, price, access
Airbnb is first because it is complex enough to be a serious benchmark and bounded enough to become a coherent independent demo.
Instagram and TikTok may be larger cultural targets, but their convincing behavior depends heavily on a live creator network, a mature recommendation graph, media pipelines, moderation, and constant content supply. A travel marketplace also depends on a network, but its core user journey can be represented truthfully with synthetic inventory: choose a destination, compare places, understand trust and price, reserve, manage the trip, and see how a host creates the supply.
Airbnb is also no longer one narrow booking flow. Its 2025 product release connected homes, services, experiences, trips, messages, profiles, host setup, reservations, calendars, and listing management. Its 2026 release added a broader travel layer, including AI-generated review highlights, listing comparison, shared itineraries, travel maps, support tools, and additional services. That makes it a compact version of the modern consumer internet: search, marketplace, identity, payments, maps, messaging, AI, trust, and operations inside one recognizable system.
Project codename
STAY/01
Independent rebuild
The working build will not use Airbnb's name, logo, proprietary assets, or visual identity. STAY/01 is a temporary HAAM project codename for an original travel marketplace that uses Airbnb as the public reference point.
This keeps the experiment focused on product architecture rather than impersonation. The article can say what is being studied while the software itself remains visibly independent.
The vertical slice
The first release should feel like one connected product, not five unrelated screens. These layers define the minimum credible loop.
Discover
Find a plausible place without fighting the interface.
- Destination, date, and guest search
- Responsive list and map view
- Useful filters with removable state
- Search recovery and flexible-date suggestions
- Saved searches and wishlists
Decide
Understand the place, tradeoffs, and full price before committing.
- Listing gallery and structured property details
- Amenities, rules, accessibility information, and location context
- Review themes with links back to source reviews
- Side-by-side saved-listing comparison
- Transparent price breakdown and cancellation terms
Book
Complete a realistic transaction without pretending real money moved.
- Availability calendar and guest validation
- Checkout sandbox with test payment states
- Identity and verification placeholders with plain-language explanations
- Confirmation, receipt, and failure recovery
- Cancellation and change-request flows
Travel
Turn a booking into a usable trip rather than a dead confirmation screen.
- Trip timeline and check-in information
- Shared itinerary and saved nearby places
- Guest-to-host messaging demo
- Contextual support with human escalation states
- Offline-friendly access to critical trip details
Host
Expose the supply-side product that guest-only clones usually ignore.
- Guided listing creation
- Image, amenity, rule, and accessibility fields
- Availability and pricing calendar
- Reservation overview and message states
- Preview of how guest-facing claims will appear
What this rebuild should change
A clone would preserve the original product's assumptions. The more useful experiment is to preserve the task while changing the priorities.
Price before persuasion
The full price and its components should remain visible while comparing options, not arrive as a late surprise after emotional commitment.
Accessibility as searchable inventory
Accessibility information should be structured, comparable, and specific enough to support a real decision rather than reduced to a vague badge.
AI that points back to evidence
Review summaries and listing comparisons should link every claim to the reviews or listing fields that produced it and remain usable when AI is unavailable.
Calmer urgency
Availability matters, but the interface does not need to turn every decision into a panic loop. Scarcity signals should be factual, contextual, and restrained.
Trip information that survives bad connectivity
Addresses, check-in instructions, host contact details, and reservation facts should remain reachable during the exact moments when a traveler may have weak data access.
Localization beyond translated strings
Date formats, currencies, address patterns, payment expectations, trust cues, and text density should adapt to the market rather than merely swapping vocabulary.
Real, simulated, and still impossible
Real in the demo
- Search, filtering, sorting, maps, responsive layouts, and account state
- A complete booking sandbox with success, failure, and recovery paths
- Saved listings, comparisons, itinerary planning, and host listing creation
- Accessibility, localization, performance measurement, and documented AI behavior
Simulated or synthetic
- Listings, hosts, guests, reviews, availability, conversations, and transactions
- Identity checks, payment authorization, refunds, and fraud decisions
- Customer support cases and host operational data
- Recommendation signals and review datasets
Not reproduced
- A global two-sided marketplace and its liquidity
- Insurance, guarantees, dispute operations, and emergency response
- Local regulation, tax collection, property verification, and trust investigations
- Years of behavioral data, global infrastructure, and a large human operations organization
The public definition of done
Complete loop
A user can discover, compare, reserve, manage, and cancel a synthetic stay.
Distinct identity
The product uses original branding, copy, components, and visual direction.
Accessible core
The main journey is keyboard-complete, screen-reader tested, and designed toward WCAG AA.
Fast enough
Representative mobile routes target a Largest Contentful Paint below 2.5 seconds in production testing.
Multilingual
The first release supports English, Estonian, and Traditional Chinese for Taiwan with layout-aware localization.
Honest AI
Summaries and comparisons show their source material, limitations, and non-AI fallback.
Transparent states
Loading, empty, partial, unavailable, error, and recovery states are designed rather than hidden.
Public accounting
The article series records dependencies, costs, design choices, failures, and what still requires human review.
Next dispatch
From brief to working search journey
The next article in this target sequence should publish the product architecture, synthetic data model, component inventory, and first interactive search-to-listing flow. Later dispatches can add booking, trip management, hosting, AI comparison, testing results, and the final accounting of what one developer actually achieved.