Hey everyone,
I initially made FlightBot as a weekend project to explore the bot space. As a pilot, I've always found it cumbersome to look up information for specific airports--it's either buried in bloated apps or on websites designed in the 1990s. I thought a chatbot would be a great medium for this kind of text-heavy information!
The bot is useful for drone pilots and non-pilots too! Drone pilots can easily look up FAA NOTAMs (important notices) around the areas they fly, or schedule a daily briefing to be texted to them. FlightBot can send you things like doppler radar (in GIF form, naturally) for your area, flight delays for major airports, webcams, or even airport restaurant recommendations.
I focused on trying to make the bot handle natural conversation as fluidly as possible and give it a little personality. You just need to give it an IATA or ICAO airport code (e.g., SFO, SEA, ORD, LHR, etc.) and it'll take care of the rest!
Would love to hear your feedback and I'll be here answering any questions!
Thanks for the hunt @__tosh!
@mironv@__tosh I tried it out and I love how it can cleverly respond to some basic conversational inputs. I'm wondering if you have any data on how often your users are trying to engage in human-like conversations with FlightBot? I'd be curious to know if early adopters are treating bots as command-line tools or as human-like agents.
@hmachalani The data surprised me a little in that I'm seeing questions with way more sentence fragments than when chatting with a real person. For example, more often people will say "weather jfk now" vs. "what's the weather at jfk?". While Messenger's Quick Replies feature is handy in that it facilitates discovery, I think it encourage a bit of this fragmented talk.
That said, when FlightBot replies with info, I see a lot more human replies like "thanks" and "this is awesome" or a follow-up question. People will also say the usual bot stuff: "I love you", "what can you do", "tell me a joke", etc. that I think is important to handle.
@mironv great analysis on how the quick replies might encourage fragmented talk. An interesting experiment might be to replace single word "cmd-line-like" quick replies with full sentence replies. For instance, if you say "Hello" to FlightBot, you'll currently get a couple of single-word commands such as "Radar", "Food" and "TAF" through the Quick Reply UX. As you called out, I think these single-word commands are teaching users to use the bot in a command-like fashion.
Maybe if you replaced those for more conversational-like inputs such as "Get me the radar for SEA" or "I need the TAF for YUL" or "What's good food near SEA" - you'll see a decrease in fragmented sentences? It's going to be harder to implement, but I'm wondering if it will directly change how users interact with your bot! If you do experiment with this, I'd love to see the results :D
@mironv awesome! Yeah I can see how that can be a challenge. Quartz has done some interesting quick replies that felt natural. It might be a good app/bot to study!
FlightBot can answer various questions around aviation. For example you can get area maps as well as information about delays at specific airports or even ask about your food options nearby. Your personal aviation assistant that is with you 24/7 and around the globe.
https://www.facebook.com/flightb...
Miron - thanks for creating a great flightbot! I have used it few times and really liked it. I have launched a digital travel assistant - Instalocate which also helps passengers in tracking flights. Please check out our PH link - https://www.producthunt.com/tech....
@ankur10 This looks awesome, great concept! Flight tracking is something I've been wanting in bot form but didn't feel like would align well with the goals of FlightBot so I'm glad someone took that on.
Replies
MOB: Mother Of all Bots
Backflip
MOB: Mother Of all Bots
Backflip
MOB: Mother Of all Bots
Backflip
Waffle
Instalocate - AI Personal Travel Assistant
MOB: Mother Of all Bots
Instalocate - AI Personal Travel Assistant