Better late than never, but I thought I’d share 20 takeaways from the points which interested me from the recent Money 2020 Asia Conference held earlier this month in Singapore. Highly recommend to anyone thinking about going next year.

Enjoy!

1. The next wave of trillion dollar companies are AI and will incorporate AI building Ai – Sentient built Google’s image classification and captioning system in 3 months – 5000 GPUs versus 30 PhD working on this for 2 years – Antoine Blondeau, Alpha Intelligence Capital
2. Using AI for automated payments fraud detection, matching round trip times from where payments directions come from and against previously held databases for users on an individual basis. Humans are very bad at making random sequences eg randomizing emails in a certain way eg name plus a sequence of five numbers – usually AI can sort these emails produced by a structured process – Michael Manapat, Stripe
3. Onboarding people without a data source (e.g passport database) by using machine learning – what does the font look like, what does the spacing look like, the paper thickness etc – Francesco Cardi, Onfido
4. Cryptocurrency shouldn’t actually be called a currency right now, you can’t actually use it to buy anything. Ripple is building systems to join and interconnect institutions – it just happens to be using the blockchain. Their goal is to take 300bp to 30bp interchange – Brad Garlinghouse, Ripple
5. SWIFT’s published error rate is 6%. As a comparison to people who say blockchain is insecure, imagine if you didn’t get 6% of your emails – Brad Garlinghouse, Ripple
6. By 2025, GPUs will be more than 1000x more powerful than comparable CPUs for AI based decisioning – Dr Ettikan Karrupiah, Nvidia
7. Computer vision will become an increasing component for investment decisoning for investment firms rather than just research eg scanning container maps for trade flows – – Dr Ettikan Karrupiah, Nvidia
8. AI is still mostly used as a facilitator – the vast majority of the ideas I saw were largely middleware, rather than taking the primary supply chain position e.g a research service that helped management consultants rather than automating management consulting
9. Real roboadvisory is not yet there because robots haven’t had to live through a Black Swan event – Augment Bank
10. Real growth in SEA payments exploding. Adyen’s recently launched omni-channel solution advertising a 6 month payback and 103% ROI. Grab’s $700m Grab Financial loan book has a data driven management systems which is showing less than 0.75% bad debts in an unbanked market.
11. Biggest question for an AI based lending platform should be how you can make cost of borrowing 5 dollars same as 500,000 dollars –  Simon Loong, Welab
12. The data shows that if you know a bad borrower on social media you are 9x likely to default than if you do not. Social analysis is not about reading your posts or judging what you look at (both illegal) – it is largely about who you know and their credit score – Simon Loong, Welab
13. When you ask people to take a photo of ID for verification purposes, the most common response is to send a picture of something else (usually a mug!) – Simon Loong, Welab
14. AI driven product creation – selling insurance for mobile phones to someone who has bought skinny jeans. Automated car damage insurance and assessment via machine learning image recognition which pays out automatically by assessing cost – Cheng Li, Ant Financial
15. Only 6% of people in Asia versus 12% in Europe versus 25% in US will not use biometrics. Corporate biometrics is 5-10 years away, currently GDPR type legislation means that you cannot use biometrics for something which doesn’t involve the protection of data and that protection is necessary (as an example in France it is currently illegal to have a biometric scan to access a door) – Philippe Biot, MeReal biometrics
16. Continuous authentication e.g how you walk, type, write etc is the future. We will be moving to only asking for authentication when there is abnormal behavior. The biggest issue which is currently not being looked at is conversely machine to machine authentication – 2 billion IoT devices to come online – Craig Ramsay, HSBC
17. Average Generation Z has 5 passwords across all their accounts, older people have at least 10-12 passwords, current generations are much more lax about security having grown up in a connected world (i.e one password gets a lot further)
18. When asked about his secret to success “Every human is built to run under 9 seconds, but only some of them work on it” – Kristo Kaarmann, Transferwise
19. Insurance for delayed flights in real time (e.g if you are late by one hour it will suggest insurance that the delay will be no more than 2 hours). If the plane is delayed payment automatically occurs via WeChat – Wayne Xu, Zhong An Insurance
20. Contrast in ways to build a hugely scalable business in Asia. Zhong An online insurance, 50% employees are engineers, no sales and no face to face – data mining only, all done via chatbot. Compare Go-pay, concept of community and principle that you can’t create trust on a phone it is all word of mouth – thousands of staff used in person for onboarding. Both have reached 150m+ users in under 5 years using very different methodologies for the same people.