Action for Children and the summer 2019 residency programme
As a leader in digital transformation Capgemini’s ambition is to help make the digital revolution an opportunity for all. In line with our mission to use our skills to benefit all society, Capgemini worked with Action for Children through a 5-week residency programme in Capgemini’s Applied Innovation Exchange (AIE), The residency was designed to help the Action for Children’s digital service team understand how innovation and technology could be better used to increase the charity’s reach and impact. The result was Elf.ai which uses artificial intelligence to support gift giving for vulnerable children.
As a result of the summer residency, Applied Innovation Team got together with Action for Children to support their Secret Santa campaign. Secret Santa is a popular way of giving presents at Christmas, but it often results in unthoughtful and ultimately unwanted gifts. Instead, this year Action for Children are giving people the option to present a gift with meaning, by visiting their Secret Santa Pop-Up Store at Covent Garden and purchasing a gift, in a unique manner, to make a vulnerable child’s Christmas a little brighter. Items available to purchase include a hot meal, warm clothes and a safe place to sleep.
To bring the magic of Christmas alive through the Secret Santa campaign this year, Action for Children told us that they wanted to host an immersive experience for shoppers, bringing to life the impact donations can have. The brief from the innovation team had a strong focus on exploring how they wanted shoppers to experience the store, as well as how they should interact with and purchase the items. Key considerations for Action for Children were;
- Creating interactive elements in the store that support the campaign and/or tangible gifts
- Learn what drives gift giving and how these insights can help future marketing endeavours
- Explore how to amplify social media posts – e.g. alongside ‘selfie spots’ that could offer supplementary digital content
- Exploring how shoppers interact with and purchase items
- Exploring how to visualise and share progress/impact (e.g. items/Secret Santas to date)
From this initial brief in early October the Applied Innovation Team went away to explore possible concepts and solutions, looking into adjacent industries and exploring the art of the possible whilst meeting the above requirements. The team spent time ideating around the problem space and came up with five concepts to playback to Action for Children for a prioritisation and selection workshop.
Five early concepts:
1. Personalised Santa
Personalised Santa shows you what you would look like as Santa, beard, belly and all, translating your face onto his body. Using a selection of photos taken in the booth, a personalised image of customers as Santa can be created using deep fake technology. Customers can get access to the image in various formats, raising A4C’s profile, while generating income.
2. Personalised Gifting
Create your own personal digital gift to send to a child in need this Christmas. Take a trip down memory lane – present a gift that you would have liked to receive as a child. Simply browse, decide and select from a variety of useful and meaningful gifts. Package your gift and write a personalised secret Santa message for a child in need.
3. Gamifying contributions
Play an immersive game where the score you reach contributes to an item needed by children around the UK – the more you play for a specific donation item the more items donated to children! This happens through a pay and play type game in which your donation gives you a number of goes at an immersive game. All this is visualised through seeing what others scores are in a leader board style with high scores and total numbers of items donated this Christmas.
4. Santa’s Secret World
Become a part of a Secret Santa world by interacting with a dynamic store window. Experience Action for Children’s campaign to be a secret Santa by personalising your own Santa Avatar in a virtual reality.
5. Thank You Billboards
Thank-you digital billboards show the generosity of those who have donated with a personal thank you message. Using the data captured through the donations process, billboards can be positioned around the pop-up store, thanking those who have donated with a personalised message.
What finally happened..
After much debate and deliberation personalised gifting and personalised Santa came out as the strongest concepts to take forward. Action for Children already had a final list of gifts that supporters could donate this Christmas. As it wasn’t going to be possible to personalise the products themselves, we needed to personalise the experience of how shoppers got to purchasing that end product.
One large factor in designing the final solution was to maintain consistency with the wider Christmas Campaign. In the campaign, anyone across the UK who donates to AFC this Christmas becomes a secret Santa and it was important for us to mirror that in our experience …
… and thus Capgemini’s Applied Innovation Exchange have created Elf.ai:
Elf.ai uses emotional recognition and artificial intelligence (AI) to help you find the perfect gift you might want to donate. To make this possible, elves present you a collection of festive photos and see how you react to them. Using AI the elves then work in the background, feeding your emotional responses into the Elf.ai prediction engine, which then suggests a gift it thinks you’d most like to donate; gifts range from a pair of warm winter shoes to a trip to a pantomime show. Once visitors have been recommended a gift they can simply pick up a tag and go to the checkout to make the purchase At the end of the experience, shoppers are transformed into a Secret Santa with a takeaway Instagram filter which they can use to spread the joy of Christmas
How we trained the model
We first created an image selection tool to get a consensus view of the types of Christmas imagery people resonate with and most importantly relate to the gifts we have available donate.
Once we had our chosen images, we then went on to train our machine learning model. We created a machine learning playground in which users were briefed on the information we were capturing through their camera and then presented several images (from our chosen selection) and asked at the end to manually select a gift they would like to donate to a vulnerable child in need this Christmas.
Whilst the experience is happening and users were presented images, we captured individual’s emotional reaction to the images using face detection, landmarking and emotion detection AI models.
With both these elements, we captured information on both the gift choice users made manually alongside the emotions they evoked whilst looking at the images. This dataset was used to train a custom machine learning engine that learnt to predict the gifts that people were most likely to gift based on these emotional reactions.
We used the Google cloud platform, React, Firebase and Google TensorFlow + TensorflowJS to create the experience and an interactive interface!
To find out more about Perform AI – Capgemini’s AI & Analytics group portfolio, please visit: https://www.capgemini.com/gb-en/service/perform-ai/
Capgemini’s Perform AI services address priority business challenges and drives impact at every level of AI maturity. From kickstarting the journey and beginning to apply AI & Analytics at scale, through embedding transformation, to realising new, disruptive opportunities. We believe in an ethical, people-first approach. AI is asking tough questions – Perform AI offers pragmatic answers, proven solutions, and real world impact. Today.
Mikey has been at Capgemini for just under two years. Based out of Capgemini Invent’s Future of Technology practice and has a keen interest in innovation and work for social good.