Covid-19 has brought unprecedented human and humanitarian challenges. Across industries, forced the adoption of new ways of working, seeing 88% of organisations adopt working from home as part of their coronavirus response plan, according to Gartner.
Our company acted swiftly to migrate to a new way of working and take learnings from the 4-day work-from-home experiment to reimagine how work is done and the role that the office plays.
Although the concept of working from home seemed to be a success for some, for others it became a challenge. The constant struggle of internet dropouts and multi-resident share houses provided an environment set for failure.
As the restrictions became to ease, we wanted to support our team in any way we could, but still maintaining an environment that was easily safe and trackable for those who entered. We also had to think beyond our staff, as they were not the only ones who entered the building, having recently opened up our building to another organisation and the flow of regular maintenance guests such as our cleaners and botanists.
All of this was a lot to comprehend and manage for our operations director. Having to move quickly, the operations and team used themselves as a communication gateway to manage who was coming in and out of the building. Team members, would text, email, slack and call the operations team to co-ordinate when they were coming in the office and how long for, before being assigned a desk and manually marked into a spreadsheet, so it could be referred back to for tracking purposes. This current mechanism became time-consuming for our Operations Director, having a non-centralised way of enabling our new team to flow in and out of the office with ease. This needed to scale and the current solution was not fit for purpose, nor scalable
As an Agency, our CEO and Operations team challenged us to with the statement of âHow might we use conversational AI to automate check-in process in order to create operational efficiencies and empower the teamâ
At first, we wanted to dive deeper into the problem, trying to understand both the core components for both our team members and the operations manager. We analysed the current workflows, marked areaâs for automaton and opportunities for cost reduction and scalability.
Instantly we recognised that we could extract all necessary information from employees, such as identify who they were, when they were checking in and where they wanted to sit, the real challenge was, how could we do this in a lean format and give transparency to our operations team. ⨠To help combat the multi-channel entry point for communication across, webchat, Slack, we needed an NLU toolkit that could enable our team to complete the check-in on the channel that suited them. In doing this we selected Googleâs Dialogflow which enabled us to expose integrations for which the team wanted to include in later phases. Additionally, to help streamline the entry point into the experience, we created a QR code, that would enable staff and our service team to initiate the initial conversation with the bot.
We also explored options for re-using existing architecture via AWS cloud formation templates to help streamline the deployment process.
To automate the desk allocation process, we allocated desks as âroomsâ through our GSuite calendar, enabling us to be able to use the Google Calendar API, as a source for managing the desks. Not only was this able to automatically book and assign a user to a desk, but it also gave us the ability to remove a required step on the frontend of the project. To manage the cap on employees that could be in the office at any given time, when the number was nearly reached, the Slack bot would advise of this within the company-wide channel to ensure that no employees wasted their time commuting to the office only to be turned down.
According to the Department of Health Victoria, itâs noted that one of the early symptoms is COVID-19 is abnormally high body temperature. As our teamâs health was a priority, we brought this into the check-in flow, using an infrared thermometer, which allowed our team to check their temperature on arrival. Parameters were set within the experience so that if an employee was 3 degrees above the standard temperature check of 36 degrees, the experience told them they were not fit to work in the office and gave them directions and suggested to get tested at the nearest testing centre.
As there were different employees working from the building each day, it was crucial to ensure that each desk was properly cleaned at the end of each day. To assist with this, the Slack bot would send an end-of-day message to all employees who had checked in, reminding them to wipe down their work area to ensure it was a safe environment for any other team member who might work there the following day.
Empowering the operations team to manage the solution moving forward was a key requirement. To combat this we created a low-fi headless content management solution via the use of google sheets API and AWSâs Lambda, enabling our team to add new desks and change content within seconds.
In the end, all of this information had to exposed to the operations team. We saved every request in AWSâs DyanmoDB but additionally exposed this via Google Sheets to make sure that it was in a familiar technology they used.
Although this experience is not the most innovative solution, its be able to provide a somewhat contactless interaction for our team to check-in and removed the average check-in/desk allocation time from 10 minutes to under 30 seconds.