reverseengineer

Thank you for both parts. Neat part: it tries harder to meet duration rather than budget, though it works to meet them both simultaneously. This is exactly the sizzle. Sydney-Hobart is 23-24h of pure driving. If you drive several hours per day it's several days. With few hours per day it's more than a week. With sightseeing it's longer. So 11 days could be exaclty optimal. Possible to tweak curiosity & comfort levels. With Long Day curiosity you will compress the duration.

Thank you, yep, we are all on pale blue dot. The plan is to cover 50+ countries, unite them into clusters, partition them into regions. Gathering feedbacks to decide what to do - add countries/clusters/regions or do GUI and usability for one country?

It caches API access token in the cookies to use for search requests, for throttling. If you search too fast, you will get human vs. robot check...

Thank you, experienced road trip professional! All points you mentioned are valid. This is only beta3 of brand new search engine, that UI is not even designed, it is very thin wrapper over bare metal AI.

We used to link cities, towns, national parks to Wikivoyage (same content as Wikitravel, but without ads). All POIs within populated places linked to Wikipedia.

You are right about visuals. Totally. I don't know where to get them from... The idea was to plug smbd's knowledge graph, with structured POIs, with curated text, pics and videos. We could easily land you by coordinates to Google Street View instead of Wikipedia. But we don't know yet what people want.

Themes is another big feature. Family or adventure or culture. Known how to do that, just not doing it right now. Thank you again for very consistent and useful feedback. This is going to impact our planning.

Those locations are definitely present in our knowledge graph. You wanted them proposed by default, without explicitly entering them? Katoomba pops up pretty frequent in our lab (but Jenolan caves don't). I promise if you specify Katoomba and Jenolan caves as start or finish or travel-through points, you will get good routes via them all.

You could enter desired sequence: Adelaide Grampians National Park Melbourne. Easy to visit Grampians in default 2 week duration. Adelaide will be start, Melbourne finish. Grampians travel-thru. But the point that Grampians had to be proposed by AI is useful. Taken.

Enter start city/town, then multiple waypoints one by one that you want to visit, and end with finish city/town. BTW it's possible to copy/paste column text, separated by new line. If you enter this sequence into that field, you will get it ready for search.

Adelaide Daylesford, Victoria Grampians National Park Melbourne

Adelaide is start, Melbourne is finish. Two waypoints in the middle, could be routed, could be skipped, depending on time, budget, comfort, curiosity levels.

Grampians National Park is present in Knowledge Graph, possible to enter as waypoint. Why it's overlooked (or bypassed)? Good question for us, will investigate... That's why showing it here!

Thank you. Possible to enter Daylesford, Victoria as waypoint. You could enter many waypoints, any duration any budget, and the route will be optimized for those desires-constraints. Maldon, Victoria is missing from our Knowledge Graph :-O I've checked it, it's cool: it has been designated "Australia's first notable town". Will improve in the future.

Agree, this is very thin wrapper on top of the tech we built. Will design UI to be both clear and engaging. Thank you for pointing to usability \m/

carlmungz, thank you for mentioning this. In UK you could start from ~20 airports. The idea is to fly to the start point, rent a car, and drive to the end point. We could increase the list of start airports to 100, will think of it. It is beta, and your feedback will shape the product.

sleiman, thank you. It recommends best possible signtseeing experience within prefered budget and time. Much more sophisticated than just routing through the cities between two points and adding POI. It is zillion times more sophisticated. I even don't understand it sometimes, then verify and validate, that thing is intelligent...

indeed, missed it, thank you, from now and on - that way.

wingerlang,

thanks for stopping by, and for your feedback.

AI because it is synthetic intelligence, it does what human intelligence cannot do. It's artificial smartness in NP complexity and CSP. You could calculate money and time expenses yourself, and figure out how accurate they are, for each request.

Yes for "geeks". Beta for travel geeks, and geeks who love to travel.