Tell us what you're building (or what keeps slipping today), and we'll propose a first sprint that delivers one complete feature you can put in front of real users fast.
Case Study
From Paper Manuals to AI Guides
Turning a fuzzy "better guest experience" idea into a shippable AI product hosts can set up in minutes.
Solo hosts and small property managers know the pattern: every new guest means the same questions on repeat—How do I get in?, Where do I park?, Where should we eat tonight?, What should we not miss?
Some cobble together PDFs, paper booklets, flyers, or long Airbnb messages; others tape a useful but basic A4 sheet to the wall and hope guests actually read it.
The result is uneven guest experience, late-night support messages, and great local knowledge that never quite makes it into a reliable system.
QR Guide is our answer to that problem, and we built it for ourselves first.
"
Guests get a PIN‑protected, step‑by‑step arrival/check‑in/checkout flow with photos and maps, plus curated local recommendations, so they don't text you "how do I get in?" at 11pm.
01
The problem
When the "guide" is a PDF, a wall sheet, or a long message thread, it fails at exactly the wrong moments.
- Guests don't have the right link handy when they land, so they default to messaging: "We're here, how do we get in?"
- Parking, access, and house rules are somewhere, but not where the guest is standing.
- Hosts have great restaurant tips and "must-do" suggestions, but they're scattered across WhatsApp, Google Maps lists, and memory, so most guests never see the best ones.
As guests ourselves, we noticed something simple: the best stays were the ones where the host made it effortless to spend our time well—with clear arrivals and genuinely useful local tips. Those were automatic 5‑star reviews. The worst experiences weren't about thread count; they were about friction and confusion.
We didn't want "a prettier PDF." We wanted a way for a host's knowledge and instructions to show up in the exact moment the guest needs them—before and during the stay, on any phone, in a way that feels personal, not generic.
"
Paper manuals and scattered messages create friction right when guests are tired, lost, or in a hurry.
02
Starting from a vague idea
The original idea was fuzzy: some mix of "digital house manual," "local guide," and "maybe a QR thing" for Airbnb/VRBO stays. We knew we wanted:
- A guide that works as a link before arrival and a QR in the property, and can also be surfaced in listings.
- A way to combine AI‑generated recommendations with host's own taste so it still feels like their city, not a generic list.
- An experience good enough that guests stop asking the same questions.
But there were real unknowns:
- Should it be "just" a web app, or something more app‑like on mobile?
- Do we start with QR printing, or keep it as a link that hosts can wrap in any QR themselves?
- How much should AI decide vs. how much should hosts curate?
Instead of answering everything upfront, we chose a narrow question to anchor the first version:
"
What is the smallest end‑to‑end product that makes arrival, check‑in, and local discovery feel noticeably smoother for both guests and hosts?
03
Our approach: a narrow, end‑to‑end first slice
We decided the first slice had to cover the whole journey:
- Before arrival: a link the host can send in booking messages so guests can explore arrival instructions and recommendations on their phone.
- During the stay: a QR code in the property (or in the listing) that opens the same guide, focused on "how do I get in / use the place / what should we do nearby?".
- During the stay: PIN‑ or token‑protected access, so door codes and lockbox instructions are not publicly visible.
On top of that, we wanted hosts to actually use it:
- A host should be able to sign up and create a usable core guide, including recommendations, in a few minutes with AI doing the heavy lifting where it can.
- The product needed a real backend API, tests, deployment guides, and a mobile‑friendly guest interface—not a prototype we're afraid to show.
"
Start with one complete guest journey—from pre‑arrival link to in‑house QR and PIN‑protected instructions—then expand.
04
What we built
The first shippable slice
On the guest side, a property guide that works as a link or QR:
- A property home screen for each property (
/p/[propertyId]) with:- Clear entry points into arrival, check‑in, house info, and checkout flows.
- Category cards for EAT / DRINK / DO near the property.
- Instruction flows for arrival/check‑in/checkout:
- Time information (check‑in from/until, check‑out time).
- Step‑by‑step guidance with text, photos, optional video, and maps, plus a "step mode" that guests can tap through.
- Recommendations browsing by category:
- Lists of places around the property with filters (price level, rating, tags, amenities, travel time).
- Map view with markers centered on the property's location.
Guests access sensitive instructions with a PIN or tokenized link tied to their reservation, so they don't need an account, but the wrong person can't see access details.
On the host side, a lightweight but complete setup flow:
- Sign up, log in, and create an account and locations.
- Add properties with:
- Arrival/check‑in/checkout texts, step lists, photos, videos, notes, and check‑in/out times.
- A chosen theme to match their brand.
- Create reservations with PINs, so each booking has its own secure access path.
- Seed recommendations using AI assistance, then curate:
- Generate structured EAT/DRINK/DO lists for a location.
- Pick and edit AI‑suggested places or add fully manual spots with their own descriptions and angles.
"
Hosts can go from zero to a branded, AI‑assisted guide in minutes—without giving up control over what guests see.
Where it is now
QR Guide has grown beyond the tiniest slice into a full product surface:
- A production‑ready backend (Node/TypeScript/Express, Prisma/Postgres, OpenAI, Mapbox) with in‑memory caching, health checks, and a broad automated test suite.
- A Next.js frontend with:
- Rich recommendation detail pages (photos, menu highlights, hours, contact details, reviews, social links).
- Admin UIs for themes, locations, properties, recommendations, reservations, and PIN verification.
- A marketing/landing page explaining the product to hosts, plus chat/AI assistant components used to help compose comprehensive check‑in/checkout information.
Hosts can now control both the operational flows (arrival/check‑in/checkout) and the experience flows (what to eat, drink, and do), all in one system.
"
QR Guide isn't a demo. It's a versioned, deployable product with real guests and properties in mind.
05
The interesting challenges
1. Making AI recommendations real and personal
Even with modern models, getting AI to steadily return real places with reliable details is not trivial. We didn't want hallucinated cafes or generic "try the local cuisine" advice.
So we combined:
- Mapbox SearchBox → AI enrichment
- Start from a real place (via
mapboxId), then use AI to enrich it into a full recommendation: long description, tags, amenities, hours, reviews, and more.
- Start from a real place (via
- Strict validation and caching:
- Use schema validation to enforce JSON shape and catch bad outputs, and cache identical AI calls for 24 hours to keep responses fast and cost predictable.
Just as important was keeping the result personal. The host's perspective has to be visible:
- The UI presents AI‑suggested candidates that hosts can select and edit, rather than auto-publishing everything.
- Hosts can add fully manual recommendations or tweak AI‑generated copy with their own "this is where I'd actually go" angles.
"
AI does the heavy lifting, but hosts stay in the loop so the guide still feels like their city.
2. Making arrival/check‑in/checkout "brainless" under pressure
The second challenge was on the UX side: guests don't read manuals for fun; they scan when they're tired, jetlagged, or standing in the rain outside a building.
We designed the flows around that:
- Step‑by‑step instructions for arrival, check‑in, and checkout, with text plus photos and maps, so guests can follow one clear step at a time instead of scrolling through a wall of text.
- Separation between public info (what's okay for anyone to see) and PIN‑protected info (access codes, lockboxes, exact entry steps per reservation).
- Host onboarding that asks only for the minimum required fields first, so they can get to a usable guide quickly and refine later.
The goal: smoother, more relaxed, trouble‑free arrivals and departures, without last‑minute calls and messages—and without compromising safety.
"
Design for the tired guest at the door, not the ideal process diagram.
06
Where we are now (and why it matters)
QR Guide is being rolled out first to our own test properties, so we can refine it against real stays before opening it wider. We treat it as a production product: versioned backend, deployment guides (Docker, nginx, systemd), health checks, and clear environment setups. It's not a shiny demo; it's a system that has to stand up to actual guests arriving at real doors.
For travel and hospitality tech founders, this matters because QR Guide is proof of method, not just marketing:
- We took a vague "make guest experience better with AI and QR codes" idea and turned it into a scoped, shippable product with a clear end‑to‑end flow.
- We built and hardened all layers ourselves: AI + maps backend, property and reservation model, mobile‑first guest UX, and admin tools hosts actually use.
- We use the same approach we bring to client work: narrow first slice, real prototype early, expand once it proves itself.
"
QR Guide proves we don't just design AI features—we build and run them in production for our own product.
07
Why this matters if you're building travel tech
If you're building travel or hospitality software and wish your product had:
- A digital guide that actually gets used.
- AI‑powered local recommendations that still feel like your city.
- PIN‑protected arrival/check‑in flows that reduce "how do we get in?" messages.
QR Guide is the blueprint for how we'd build it with you.
If you're sitting on a vague AI idea—"something with AI around guests, guides, or recommendations"—or you already know the guest experience feature you wish your product had, tell us about it. We'll propose a first 2–6 week sprint that gets you from fuzzy idea to a production‑ready slice, with a working prototype by Week 2.
"
Tell us what you're trying to build for your guests, and we'll map a sprint that turns it into a working feature—fast.