Close to home
What does the world’s largest online retailer, and Australia’s largest health insurer have in common?
They’re hoping to grow their (already substantial) market share by getting closer to their customers, literally. Amazon is building out hundreds of ‘last mile’ delivery stations, in an effort to make same day delivery the default for more people. And Medibank is also trialling a ‘last mile’ approach: by geo-caching inbound enquiries to its call centres and getting local people to answer those calls.
Why?
Because if you live in Geelong, with a quarter of a million people, a Geelong local will tell you much more about the health system where you live than someone in Maroochydore, or Bangalore. The idea is that ‘triple level routing’ will enable a call to be sent first to a local, then a state, then a national operator, depending on how busy they are. Medibank is trialling at two scales: Geelong’s customers will have calls answered by its Geelong store staff; calls at an entire state level, South Australia (population 2 million), will go to its Adelaide stores.
The aim is to ‘decomplexify’ an often confusing system, by giving people very concrete information, using local knowledge, and short-cuts. And, early trials show that people find people who live close to them, 'relatable’.
Their measure of success? It’s quite simple: Customer satisfaction (which took a hit during Medibank’s data breaches last year).
Question: How can you scale, by going local?
The end of generic information
I was an early adopter of ChatGPT and almost daily push at the limits of what I ask it to do.
For a client project, I got it to nail down the essential differences between ‘health literacy’ and ‘patient activation’. My client, who’s an expert in precisely that field, said, “That’s a better description than I could have written”.
I’ve shown my good friends Louise and Lachlan how to get it to write compelling descriptions of antique jewellery for their wonderful shop. And, I routinely get it to explain basic concepts to me whenever I start working in an unfamiliar field: The history (and future prospects) of REITs, anyone? When to use in-vitro vs in-vivo testing? Or, the difference between climate change adaptation, and climate change mitigation?
Now, all of the above required me to get good at queries. Ask ChatGPT a rubbish question, you’ll get rubbish answers. But, sniffing the AI winds, I sense that’s all that’s about to change.
Literally, this week.
As of this today, any of us will can to get ChatGPT to create our own GPTs. Without help from additional software or any coding knowledge.
In other words, my client, the UN Environment Program, can rapidly create an ‘Environment GPT’ that answers any environmental question, from any person. My friends Louise and Lach can build a ‘Jewellery GPT’ that answers any question a customer might have about 19th century mourning jewellery, or why opals are a mineral and not a crystal.
Needless to say, this hyper-contextual responding will be huge.
Imagine a university creating this for every subject it offers: students can contextually ‘talk’ to the bot to learn. Imagine a hotel or Airbnb offering this to help guests get to know the nuances of a local area. Imagine an architectural or interior design firm creating a GPT of all of its prior designs and letting customers ask questions about room dimensions, colour schemes, or light.
It’ll be like the Apple App Store (which has generated over one trillion dollars in sales since its inception. Yes, that’s a ‘t’). Oh wait, OpenAI are launching a GPT Store. Like the App Store, it will revenue-share with you or I, should we wish for our GPTs to be available to the public. I can’t wait to see what people’s ingenuity will create here.
Question: How could you use ChatGPT to deliver personalised information to your customers or clients?
Extrapolation
I spend my professional life making distinctions: categorising and organising logical levels of ideas; separating out similar concepts; unifying dissimilar concepts.
I’ve long loved the adage: “There are two types of people in the world. Those who think there are two types of people. And the others”.
So, I did a double-take when I saw a fruit vendor here in Nairobi, with this t-shirt. I was in a car, and couldn’t grab a photo, but I did find it online.
I’ve spent the week with environmental scientists, who are truly expert at extrapolating (Definition: “Extend the application of (a method or conclusion) to an unknown situation by assuming that existing trends will continue or similar methods will be applicable).
They’re very good at extrapolating across materials (“If this strategy works to eliminate plastics, could it work to eliminate hazardous chemicals?”) as well as extrapolating across settings (“Can we use the same incentives to get African villagers to plant trees to get Chinese villagers to plant bamboo?”). And, of course, extrapolating across multiple boundaries (“If we eliminate toxic chemicals from clothing, we make it more biodegradeable when it becomes waste, and it protects human health, during production, use and disposal”).
So I loved this. I want one.
Question: How can you extrapolate to create innovative solutions to difficult problems?
As you read this, I’m heading from the sub-tropics of Kenya to the almost-winter chill of Geneva, Switzerland. And, because I know a few of you will ask, here’s a gratuitous tourist snap. It was truly an incredible experience being up SO close to such enormous creatures - this is Daisy, a 20yo female, whose head is the size of my torso.
See you next week,
Andrew
Loved the hyper-local concept - though it does seem ironic that it's taken so long for it to be (re)adopted! Local knowledge has always been simply invaluable...
And the AI / GPT developments are mind-bogglingly exciting!
Great read and oh wow Andrew, you and Daisy, just beautiful. Safe travels