Conference Day Two: Wednesday, 27 September 2017

8:00 AM - 9:00 AM Coffee & Registration

9:00 AM - 9:10 AM Opening Remarks from Conference Chairperson


9:10 AM - 9:40 AM How to: Work With Bots: Developing, ‘Upskilling’ and Learning From Mistakes

Magda Cortez, Senior Product Manager, eBay (Gumtree)
Gumtree, one of Australia’s top 12 websites, has been developing, fine-tuning and Upskilling its ‘Gumbot’ and Facebook Messenger Bot for the past 2 years, using them internally and externally to boost CX. They are now at the stage of trying to add a more personal touch, giving the bots personas and personalities.

Key Question: How do you go about rolling out Bots and ensuring the most accuracy and least amount of disruption to customer service?
Focus: Bots, MVP, getting the balance right, Upskilling bots, benchmarking, customer-centricity, FAQs, categorising responses
Biggest Hurdle Overcome: Continuously reprogramming and 'upskilling' the Bots

Magda Cortez

Senior Product Manager
eBay (Gumtree)

9:40 AM - 10:10 AM How To: Design an Intelligent Agent to Help Drive Solutions and Profit

Key Question: How do you develop effective algorithms to maximise profit margins?
Focus: Intelligent agents, algorithms, ROI, solutions, increased profit margins
Biggest Hurdle Overcome:

10:10 AM - 10:40 AM Solution Provider Session

10:40 AM - 11:10 AM Morning Tea

11:10 AM - 11:40 AM How To: Improve Customer Service Through Intelligent Virtual Assistants

William Yeoh, Director for IBM Centre of Excellence in Business Analytics, Deakin University
Deakin University was a pioneer in its use of IBM Watson, rolling out Natural Language Processing across the organisation. Now it has moved onto Genie, a mobile app platform made up of chatbots, artificial intelligence, voice recognition, and a predictive analytics engine. Students are now able to access everything they need from an intelligent virtual assistant.

Key Question: How do you improve customer service through AI and ensure it is rolled out seamlessly with minimal disruption to BAU?
Focus: IBM Watson, Genie, mobile applications, intelligent virtual assistant, customer service
Biggest Hurdle Overcome: Learning from mistakes along the way and ensuring they aren’t repeated

William Yeoh

Director for IBM Centre of Excellence in Business Analytics
Deakin University

11:40 AM - 12:10 PM Solution Provider Session

12:10 PM - 12:40 PM How To: Use AI to Manage Different Customer Cohorts and Improve Customer Service

Tabcorp is improving automation around its customer contact and management strategy, looking at how it manages different customer types and how to deal with specific cohorts. It is also using AI to inform better pricing using sophisticated data analysis.

Key Question: How do you develop intelligent algorithms to manage different customer types and provide a more personalized service?
Focus: Customer service, CX, identifying customer trends and those at risk, intelligent algorithms, pricing
Biggest Hurdle Overcome:

12:40 PM - 1:10 PM How To: Improve Customer Experience Using AI Self-Service Capabilities

Vodafone NZ has implemented an iterative program using AI around service experiences to help guide customers to self service answers and specific information for their queries. The organisation has a lot of different systems and complexity, and 20-24 contact numbers for customers to filter through. Rather than wasting time and money, they have used AI to quickly deliver proof of concepts and gain real runs on board for CX and cost savings.

Key Question: How do you improve CX using AI?
Focus: Self service, cost reduction, deflection of cost savings, improved accuracy, learning from other companies’ mistakes, starting small and iterating
Biggest Hurdle Overcome:

1:10 PM - 2:10 PM Lunch

2:10 PM - 2:40 PM How To: Use Image Recognition to Cut Cost, Improve Accuracy and Increase Coverage

Agustinus Nalwan, AI & Machine Learning Technical Lead, is using image recognition technology to minimise cost, boost accuracy and increase coverage. Until now, every image uploaded by their photographers (20,000 photos a day) was manually categorised according to the angle featured in the image. User uploaded images (80,000 photos a day) didn’t receive the same attention, but now both streams are done automatically, with image comparison tools now being developed.

Key Question: How do you develop AI capabilities in house and upscale the project?
Focus: Image recognition, DIY, open source library, training and continuously teaching a system, figuring out how it learns, training strategy, indexing images
Results to Date:
· Cost reduction for just the car vertical: 15% (25k a month)
· Accuracy: 97.2% improvement
· Coverage: Double the photo indexing coverage to include private seller car photos
Targets: Build several consumer facing products such as photo comparison tools (compare dashboard/boot photos of cars) and private seller photo advisory tools
Biggest Hurdle Overcome: Data preparation and then training strategy

Agustinus Nalwan

AI & Machine Learning Technical Lead

2:40 PM - 3:10 PM Debate: AI Will Provide a Better Service Than Humans to Customers

AI, once refined, may be more accurate, but is it more personal? Will it provide a better service to customer and can we rely solely on AI as the interface for 100% of our customer service, or does there need to be a balance? Then there’s the factor of emotion and empathy – will AI advance to the point it is as good as/better than human contact – if so, when?

3:10 PM - 3:40 PM How To: Bridge the Gap between RPA and Intelligent Automation/AI

Having mastered RPA, which has freed up staff at Westpac NZ for higher value tasks, they’re now looking to the next phase which will be the implementation of intelligent automation and AI applications. RPA has put all the parameters in place for things like access, how it affects HR and reporting lines, security, risk identification etc, making AI a lot easier to navigate.

Key Question: How does RPA set a solid foundation for Intelligent Automation/AI, and how do you leverage it to take that next step with ease?
Focus: RPA, internal and external chatbots, process efficiency, data generation and utilisation
Results to Date:
Process efficiency has improved by up to 80%
•SLA times have dropped from 5 days to just 1 day
•Data entry and risk assessment accuracy has improved
Improve efficiency and risk reduction
•Expand to pan-bank automation
•Improve customer service and turn around times
Biggest Hurdle Overcome: Governance and risk acceptance to test in production, ie. Treat RPA as a first time user in training for testing phase and fix inaccuracy as you go through the process.

3:40 PM - 3:50 PM Closing Remarks from Conference Chairperson

3:50 PM - 3:50 PM Afternoon Tea and Close of Conference