How Tabi compares, what it costs, and how the library gets built.

WanderLog is a shared trip planner: strong at pinning places on a map and splitting bills, weak at telling you what to actually do. You still start from a blank canvas and paste in ideas from ten browser tabs. Tabi starts the other way around: describe your trip in a sentence and Tabi drafts a full day-by-day itinerary from a curated library, with real opening hours, walk times, and booking links already stitched in. You can still edit and share like WanderLog, but you're editing a plan rather than building one from scratch. Think of WanderLog as the spreadsheet and Tabi as the friend who already made the reservation.

Wonderplan and Layla are chat-first: you volley messages with an AI and hope the itinerary sticks. Curiosio optimises road-trip routes and Kayak AI bolts a chatbot onto flight search. All four hand you a draft, then leave you to book across five other sites. Tabi treats the itinerary as a living document you keep, not a chat log you lose: every activity has a verified photo, live opening hours, a Google Maps deep link, and a bookable partner (GetYourGuide, Booking, Skyscanner, Duffel, Airalo) attached at the right moment. Plan once, book from the same page, keep the trip forever.

You can, and the first draft looks great. Then you notice the museum it recommended closed in 2019, the restaurant address is wrong, and the 'ten minute walk' is actually forty. General LLMs hallucinate specifics because they weren't built for travel and have no live data. Tabi uses the same class of model but grounds every place in Google Places (verified photos, opening hours, coordinates), matches long-distance legs to real transport APIs, and asks a second model to fact-check before publish. When Tabi doesn't know, it says so, instead of inventing a plausible-sounding lie.

The free tier is genuinely free, no credit card needed. You get up to 5 trips a day (100 a month), saving and sharing, and full read access to the curated library. If you go over the daily allowance, it resets the next day. Pro (USD 4.80/month or 58/year) lifts the cap and adds features like calendar sync and group expense splitting. No tokens to count, no ads, and we don't train on your trip data.

A Google search sends you into a Kayak or Booking.com results page where you re-enter your dates and destination. Tabi's booking links are pre-filled: click the Ubud yoga class in your itinerary and GetYourGuide opens on that exact operator, click the Kyoto hotel and Booking.com opens on that property for your dates. Behind the scenes we use each partner's official affiliate deep-link (Travelpayouts-brokered where required) so pricing is identical to going direct. You do not pay a Tabi markup; the partner shares a small commission with us when you book, which is how the free tier stays free.

No. Tabi runs on subscription and affiliate commissions, so we never need to sell trip data or profile you for ad networks. Trips you save are stored in your account on our own self-hosted Supabase (EU-region), covered by RLS so nobody, including Tabi engineers on day-to-day work, sees another user's plans. We do not sync trip data to Google, Meta, TikTok, or any advertiser. Analytics are aggregated and privacy-first (Datadog RUM, no third-party ad SDK). You can export or hard-delete your account any time from Settings; deletion is irreversible within seven days.

It happens, less than you'd think, but it happens. Every activity Tabi surfaces has a thumbs-up / thumbs-down button; a thumbs-down sends the place into a review queue and drops it out of future itineraries within the hour. We also cross-check landmarks against Google Places at generate time, so if a spot doesn't resolve to a real business with recent photos, it never reaches your itinerary. Compared to ChatGPT (where a wrong museum stays wrong forever) or WanderLog (where you have to fact-check yourself), Tabi's plan gets more accurate the more travellers use it.

Yes, on the mobile app. Once a trip is saved, the itinerary, maps, photos, and booking confirmations cache locally so you can open the day's plan on a rural train or a Balinese beach with no signal. Offline mode also stashes your PDF ticket, the address in local script, and a walking-directions snapshot for each activity. What you can't do offline is regenerate the itinerary or search for new places, since those need our servers. Web browsers get a lighter version: recently viewed trips stay in cache but a hard refresh needs network.

Every trip in /curated is drafted by our authoring model against a research brief for that destination, then run through a separate reviewer model that fact-checks landmark names, opening hours, and any specific claim (prices, distances, seasons). Anything the reviewer flags as fabricated or stale is either fixed or the whole draft is rejected. A human editor signs off before publish. The library grows about ten destinations a week and every trip is versioned, so if a museum closes or a ferry route changes, the next tick corrects it and you see the update on your next visit.