Making cities talk. Smart city chatbots with the City of Adelaide!

An inside look into executing a chatbot project with a local government.

In a hyper-urbanised world, smart cities are adopting technology to optimise citizen support & improve accessibility to urban services. To achieve these 2 goals, Hopstay & the City of Adelaide launched a project to integrate chatbot technology across the City of Adelaide organisation.

The City of Adelaide is the central local government area for the Adelaide metropolitan region, Australia’s 5th largest urban area & the capital city of South Australia.

The chatbot (and now voicebot!) allows citizens to undertake basic citizen services & access key city-based content through a chatbot residing on Facebook Messenger & as an integration into the new City of Adelaide website.

Here’s how it went down!

Step 1, automate responses to hundreds of FAQ’s

Councils receive repetitive inquiries all day every day.

• When is my bin being collected?
• How do I pay my parking fine?
• When are my council rates due?

Just to name a few.

Integrating a circle of automation around their organisation allows City of Adelaide customer service agents to focus on complex inquiries & speed up response time. It also ensures that rate-payer dollars are spent on citizen-led projects, not a growing customer support burden.

After an initial import of FAQ’s from their pre-existing knowledge base, Hopstay’s natural language processing modules got to work to ensure that questions were understood & responded to correctly. This continues to evolve with time, thanks to new data coming in every day.

Step 2, allow citizens to report issues in their neighbourhood via the chatbot

Reporting an issue in the neighbourhood is a commonly requested council service. Citizens frequently report things such as:

• Graffiti
• Cracks & potholes
• Rubbish
• Faulty parking machines

Etc etc etc…

Making use of Facebook Messenger’s inbuilt camera & location sharing, we were able to provide a mobile-friendly way for citizens to report an issue without requiring them to download another app (given 81% of Australian internet users (those ages 16–64) already use messenger apps & Facebook Messenger has the highest prevalence).

They can write notes, take a photo of the issue & enter an address without leaving the Facebook Messenger environment.

The module required an integration with City of Adelaide’s technical systems to ensure that the issues, once received, were triaged to the relevant department.

Step 3, solve the bin question!

As stated before, when is my bin being collected is one of the most common questions coming through council customer service channels.

To solve this, we took the waste calendar & mapped out the geography so we knew the bin day of every address within the City of Adelaide. We set up an API so that every time a user asks this questions, the chatbots asks them for their address. After receiving the address, the chatbot returns the upcoming collection dates.

Then, making use of Facebook Messenger’s broadcasting infrastructure, the chatbot offers to send the user a notification every time their bin is being collected.

Hundreds of Adelaide citizens will never miss bin day again!

Step 4 — now that we’ve solved bin day, it’s time to work on waste disposal

How many have you stood in front of bins trying to work out which bin to put item X in to?

Ha, knew it!

This module enables users to type in an item & receive a response from the chatbot telling them which bin said item belongs in.

This implementation required a large database of common household waste items, something that the City of Adelaide worked hard to provide. Once we had a database large enough (approximately 150–200 items), we had to identify all of the different ways a user might refer to that item. This included training the bot to understand synonyms, word order, different types/models of certain items & of course (as always) spelling errors.

We now have a fast growing database of items. And the City of Adelaide have agreed to share this database with other councils. We’re looking forward to rolling it out in other cities & building on the dataset!

Step 5, parking

Creating citizen-centric use cases out of available urban data is a core objective for both Hopstay & the City of Adelaide moving forward. Our smart parking module is a good example of this.

This integration required us to connect live parking data from the UPark parking centres dotted across the City of Adelaide. The information is then displayed live in the chatbot on request of the user.

Creating citizen-centric use cases out of available urban data is a core objective for both Hopstay & the City of Adelaide moving forward.

This helps the user plan their trip & search for available car parks near to their destination. It is a particularly useful implementation for CBD council areas with high parking demand & constant inflow & outflow of workers & commuters.

[Our next step is to roll this out to on-street parking — all in good time!]

Step 6, our first foray into voice

Hey Google, can you ask Adelaide when my bin is being collected?

For cities, accessibility is paramount. This is due to their responsibility as public bodies to to ensure all citizens can access services.

More than 1 billion people around the world now have access to voice-activated assistants, and they’re a great tool for disabled people & the elderly who may lack the capacity to visit a customer service centre or make a phone call.

Google Assistant was our platform of choice due to it’s prevalence in Australia. By saying ‘Hey Google, speak to Adelaide’, the user triggers the City of Adelaide voicebot, which provides urban services via smart speakers & voice assistants.

So far, check bin day is the only module we’ve deployed to Google Assistant. We’re working on many more & look forward to launching increased voice functionality over the second half of 2020.

Step 7, review, analyse & accelerate

With more than 8,500 users & having received more than 30,000 messages, we now have enough data & feedback to accelerate development where we see fit.

We are currently designing additional functionality, new use cases & strengthening the intelligence level of both the chatbot & the voicebot, based on the following key takeaways:

The case for optimising customer support is strong

Almost all urban areas are growing. Some more than others. Even improvements of 20-30% in processing customer support inquiries can have significant organisation-wide impact.

The case for accessibility is strong

Consumer trends are evolving. Consumer expectations are changing. Different platforms are emerging. An accessible organisation is an organisation that understands their users’ preferred interaction channels.

Urban data is a big opportunity

It’s everywhere & cities are increasingly making it public. But, a lot of it is still unused. There are an enormous number of potential use cases waiting to be developed.

For more information on this project & access to the full case study, please contact lucas@hopstay.co.