💡 How to manage shift attendance with Make and ChatGPT

Hey Makers :wave:

AI is a hot topic these days, and its potential to simplify our lives seems limitless.
Today, we’re chatting with Angus Gastle, President of Phosphene, about combining AI and Make to effectively manage shift attendance. :man_office_worker::bar_chart:

:mantelpiece_clock: Taking a look at the Canadian labor market, we see that an estimated 12% of working adults are engaged in shift work. With so many people working variable shifts, this represents a significant scheduling challenge.

:brain: No-shows and accountability issues were a particular problem for Phosphene’s client, a security company. As it turned out, all they needed was a little bit of Make and AI magic.

With ChatGPT, the company is now able to text employees to confirm their availability and then track their responses, allowing them to confirm attendance or schedule replacements as needed.

:muscle: The results include reducing no-shows to almost zero and managing up to 20,000 shift hours seamlessly.

Read on to learn how they did it!

What problem were you trying to solve with your automation?

Reducing Shift No-Shows for a large Security Company here in Canada, using ChatGPT and their Scheduling Software (TCP Humanity).

Why did the problem exist?

When you’re running shifts at scale with casual staff, there can be some accountability issues/churn for folks who have last minute schedule changes or otherwise decide they can’t make their scheduled shift. Shift Supervisors may not find out until 10-15 minutes after the start of a shift at a large event that someone has not arrived on schedule. This means they need to find their contact details, and call them to confirm their status - and possibly try to find a replacement on short notice. Our goal was to reduce or eliminate the need for this last minute scheduling issue by confirming just prior to the start of the shift everyone’s status as a double check that there will be 100% attendance.

Scope: Every month this company is tracking between 6,000 to 20,000 recorded shift hours depending on the time of year. That’s a lot of shifts and team members to keep track of!

Who needed it: A Canadian security staffing company with a large casual worker pool.

How did you solve the problem? What does your solution look like?

Our solution is delivered in two parts. The first scenario is the Initiator, which looks for imminent, upcoming shifts and texts each member on the team to confirm their status using natural language generated by ChatGPT (Now using GPT-4).

The second scenario is tracking Twilio and reacting to all inbound text messages, adding it to the conversation history in a data store, and then using ChatGPT to both decide their intent (On time, Late, Missing or Other) and take an action within the scheduling software to Confirm or Cancel their shift assignment. This leverages the Scheduling Software’s API via an HTTP module to take these custom actions based on how ChatGPT interprets their arrival probability from their response.

The result

The automation behind the scenes

What did your solution achieve?

By integrating this automation series within the company, we’ve been able to drop No-Shows to almost 0 ahead of the busy season this summer. We’ll be tracking progress to see how well this performs over the busy months.

Reducing last minute panic Reducing confusion for Shift Supervisors Increasing the billable hours tracked across all shifts

Angus - the man behind the automation wheel

Helpful Resources:

:make: Twilio on Make
:make: OpenAI (DALL-E and ChatGPT) on Make
:make: 🟣 Getting started with OpenAI GPT-3 and Make
:make: Blog: How to Automate your Workforce with AI and Automation
:make: Blog: 5 Complementary AI Skills to Survive and Thrive in the AI Era