Voice assistant technology holds a great amount of potential for spatial querying. In this article, Qwhery CEO Matt Pietryszyn discusses his smart speaker application Q11, that uses voice assistant technology to answer geospatial data queries, which new features are on the company’s roadmap and shares some best practices for upcoming entrepreneurs who want to start their own business.
Founded in 2019, Qwhery combines voice technology with geospatial data queries, in order to streamline interactions between citizens and their municipalities. One of the company’s co-founders, Matt Pietryszyn, worked almost 20 years in the geospatial industry, building applications using spatial technology and open data with large teams across municipalities.
Qwhery’s Q11 smart speaker application and query cloud
During this time, he recognized a strain on 311 resources and developed Q11, a smart speaker application that connects smart home devices to local government Open Data, and taps into information about services and local events or to report problems. Originally, Pietryszyn saw the concept of using voice technology being applied in John’s Creek, Georgia, where a an application was built that coupled voice technology with GIS data and technology, and that enabled citizens to use their Alexa-enabled devices to ask questions and get responses from their city.
Pietryszyn took this idea and scaled it up to build a platform that many municipalities could use to integrate their GIS with, in addition to their open data. “I thought it was really intriguing and something I could do relatively quickly, to bring some new modern technologies to citizens and have municipalities do something innovative”, he says.
Matt thought of a way to extend this capability and tap into the service information that cities have and that people are looking for. Examples of these are people asking when their garbage will be picked up, why it hasn’t been picked up yet, or what’s being built within five kilometers of someone’s house.
Query cloud, allows municipalities to integrate their GIS systems with web services or APIs and Build custom voice applications. For Amazon Alexa, these are called skills, while for the Google assistant framework, they’re called actions. Qwhery has built their own skill and action for both of those brands and named it Q11.
The name Q11 is a take on municipalities call center and the phone number 311 that a resident would typically call to speak to a call center representative and get information about their municipality, their home and property. Q11 is available on each platform’s marketplace so that a user can engage with that voice application on whatever device they’re most comfortable with. It connects to the query cloud, which then connects to the individual municipalities’ business system and provides that information back to the user.
Qwhery built a single app that does many things, enabling each municipality to integrate what they feel is appropriate for their residents:
“A municipality could pick and choose which business systems or data endpoints whey would like to connect to it. For example, they could integrate information about their members or council, their wards, districts or an event calendar. When users interact with Q11, it uses their location of where they live using their device to understand their location and which municipality to connect to.”
Turning spoken user requests into logical statements
Q11 is able to turn a user request into a logical statement that can be executed against a database. This is possible because both Google and Amazon use natural language processing (NLP) in their voice platforms, which takes the words that people speak, ingests them and transforms them into text and information that can be turned into data and passed forward to other APIs in order to query information back.
Pietryszyn states that Qwhery created synonyms for different phrases to handle variations of what a user might say to train the model in Google and Amazon Alexa so that they know that any sort of form a certain phrase means this one, specific thing. For example, when a user asks for all construction projects nearby, the word “nearby” is set to a variable in the query cloud framework to 2,5 km. That parameter is passed to the ArcGIS Feature Service API with some more data, so that Q11 is able to pull back all the construction projects within 2.5 kms of the requester’s current location.
Natural Language processing and geospatial use cases
To be able to provide contextual information back, Pietryszyn says that some understanding is required of the region in question. For example, the US uses districts, where multiple members of council represent single districts, whereas in Canada the concept of a municipality is broken into geographical areas called wards, where each wards has a member of council: “To better understand the region where working with, we provide a municipal administrator some options for regional differences they can select when they log on to the query cloud to hook their REST endpoints into the query cloud so that our skills and actions can work with it. Those options then become parameters within the query that we use and they message that the user receives back from the Google Assistant or Alexa”, he explains.
One example of a query that would make a lot of sense to an individual about their property might be what is the zoning for one’s property or what are the permitted uses for one’s property, says Pietryszyn: “that information can be really difficult to find for anyone who goes to a city’s website and tries to look it up. We’re able to provide that information back instantly for your property”.
Another interesting concept is just simply providing information about what’s happening around people. “Especially after being locked up for two years during the pandemic, so many people are going to be ready to head out, into their communities and looking for interesting things to do. I think that’s something that a lot of people would get a lot of use out of”.
Future Q11 product features
Apart from citizens asking questions and getting back answers, Qwhery is already working on new product features such as sending push notifications from smart devices. The first use case is to push out reminders to residents when a garbage truck is on their road or heading their direction and to make sure they get their bins out to the road, or even going so far as to reminding them which week is recycling week and make sure they get their green bins out for organic waste.
Apart from reading from a database, Qwhery is working on an application that writes to a database. “We’re currently hooking into one of our customers’ database through the ArcGIS REST endpoint. They’re letting their residents report around 50 different deficiencies through voice technology. All that information is recorded through that feature service REST endpoint into their database.” What’s interesting is that a lot of the fields in the form can be entered automatically based on the user’s Google or Amazon profile information, including their location, phone number and email address. The forms Qwhery is building today have some free form where it captures the user’s input as a large text blob and stores that in the database, but the company also provides a set of options to choose from by saying “select” or “next” to move to the next option.
Open source integration and possible competitors
Qwhery connected with the ArcGIS platform because Qwhery is an emerging partner with Esri, so that was where the company focused their efforts first. Pietryszyn states that there’s nothing that would stop them from scaling that integration or extending to an open GIS platform or anything that has a GeoJSON endpoint: “We pass information to Google geocoders, we can read and write to XML, we are integrated with the general transit feeds specification (GTFS), so we definitely integrate with plenty of different platforms, whether it’s COTS GIS or free and open source, or plenty of other business systems”.
When asked if he sees both Google and Amazon as possible future competitors, Pietryszyn points at the open data journey that municipalities have been on for many years now, without the emergence of a standard for zoning information, wards and members of council information service request information. He says he doesn’t worry too much that Amazon or Google will be able to consume all of this information that’s shared openly by municipalities because there is no standard: “What we’ve built with the query cloud is that integration point that cities can connect to, and it doesn’t matter if they’re standardized or not. They connect their REST endpoints and map to their data spec on the fly”, he adds.
Creating awareness to enable reusability
Pietryszyn states that the biggest challenge his company faces now is to make things sticky and reusable, meaning getting people to know that this is out there and to use it: “Because a question such as ‘what’s the zoning for my property?’ is not going to be asked every day, week or month, but only once, the biggest challenge then becomes to get the awareness that this capability exists and can be used for many different things so that people come back and use it often and make it part of they interact with their municipality”.
Initially, Qwhery’s main goal was to try and help municipalities to lower the call volumes they receive at their call centers and shorten the wait times for citizens. During the pandemic, where people didn’t have the ability to show up at city hall, a lot of that service moved either online or to the phone, so Qwhery’s goal was to create this other channel to interact and communicate back and forth between the city and the residents.
The future of voice technology as an interface
Looking at the future developments of voice technology, Pietryszyn thinks the world is at the beginning of using voice technology as an interface. He expects that voice technology will become really prevalent as the world moves into more and more screenless devices. As an example, he mentions diving into the metaverse, where people are going to need to use their voice more now that menu systems aren’t the same with using a mouse to select your input.
He points out the benefits of voice technology for specific queries compared to web mapping: “All municipalities have tons of web maps that they share with their community members. It has taken a lot of time to build up the adoption and awareness that they’re out there and get people to use them. But everybody has a voice enabled device, which I think is something they’re more comfortable with than having to go and learn how to use a web map for find one, as it can take a long time to find that information before you can use it. Using voice is quick and simple, it’s just a matter of asking a question and getting a response back”.
However, he admits that voice technology is not always the right approach, and especially not when trying to replicate the front-end experience of a large, complicated menu system: “First of all, it gets really difficult to first program and train a model for that, but also for the end user to interact with. We came up against this last year, where we were trying to develop a solution with voice that had a very complicated and large menu system that users had to navigate on screen. Halfway through our first approach of trying to replicate that, we realized it just wasn’t working the same way. So we had to re-engineer the way that somebody would interact with this information using their voice.”
The future of geospatial voice interfaces and some advice to future entrepreneurs
Qwhery is now a startup, meaning it’s an evenings and weekend project for its founders. The company started as a passion project, as Pietryszyn wanted to do something technical and innovative: “Looking for new ways of doing things has always been my sports. My hope is definitely to grow the business and make it more of a fulltime venture”, he says.
Pietryszyn has some advice for people in the similar situation as him: “What I can say is that if you have the passion and tenacity to keep going with something you’ll know when it’s time to go fulltime or that something will happen, such as getting some sort of breakthrough contract where you’re going to need more time than you put in in evenings and weekends. That large contract will be a big jumping point from which things will spiral upwards. For us, we know what we’re aiming towards, but as you grow your company and bring in more staff members and collaborators, that goal is moving too, because you need to be able to responsibly fund the company and be able to pay staff and all those other things go along with running a business”.
To someone looking to start a business, Pietryszyn’s advice would be to start, build a business plan and think about a network because as starting entrepreneurs will rely on that network. “Also, get yourself a mentor to share your idea with and trust, someone who started a business. I did this in 2019 and had lunch with an industry contact who started a geospatial business and had a lot of success. I invited this person for lunch and asked as many questions as I could. It’s really important to hear their information about building a business and understand that it’s going to take a lot of your time and to be able to push through a lot of rejection, but to keep going with it if you believe in your idea”, he concludes.