AI-specific examples Flashcards
What types of ML algorithms did your teams use?
Depended on the data sets - whether they were structured or not, labeled or not, and what we were trying to do.
e.g. we had classification models including both decision trees and k-means for triaging and assessing whether rehab was relevant. Our NLP models were built by a vendor
What were some of your key learnings building AI models?
- The leap between prototyping and productionisable models for us was huge
- Assessing the availability of production data and having infrastructure to support models needs to be done up front
- Change management - people really don’t understand the what’s possible and have fairly obscure ideas about what AI is
How do you go about developing a product roadmap?
Step 1: Product Strategy: Deeply understand customer needs, stakeholder needs and company objectives. Clearly articulate the product strategy and get agreement on that strategy.
Step 2: Draft & Collaborate: Build first draft using high level milestones and collaborate (dates, dependencies, capacity, etc) - 1:1 meetings with stakeholders to get feedback on the draft. Point of roadmap is to align stakeholders and set direction. Examples of not enough time on collaboration and milestone confirmation.
Step 3: Roll out: and ensure every stakeholder that needs to know, knows. It then forms part of your quarterly OKR’s.
Ongoing: Maintain/keep relevant - dynamic documents. Reprioritise only if something changes - customer needs, competitor info, etc.
Example - roadmap for claims incl. customer portal MVP, various AI integrations, automated SMS follow ups, rehab support, etc.
What are some of your techniques for getting customer feedback?
NPS, surveys, roundtables, customer panels, ethnographic.
Example - level premium misunderstanding
What were the AI use cases you built?
Objectives: efficiency (CX, opex), accuracy, RTW
Models: Mostly focused on income protection initially. Highly complex because we’re assessing 1) that people are unable to work according to their policy definition, 2) their pre-disability income and therefore the benefit calculation. Also, triage & RTW. Walk through process and prioritisation of models
What excites you about the Search and Recommendations space?
I think recommendations in particular is interesting. Especially if external data is being used… I think there’s huge potential to present customers with opportunities they wouldn’t have necessarily considered.
Why should we hire you for this role?
- I’m a shaper in that I can work across developing the strategy, the vision and roadmap, and also lead teams to execution.
- I’m good at articulating complex concepts in a way that gains buy-in and have done this at all levels
- I have built a customer journey using AI and have experience in all of the change management and lack of understanding that comes with it
Why are you interested in this role?
Role: I like that it has responsibility for the product strategy, roadmap and implementation of AI.
Product: For me at this stage it’s more about solving customer problems and delivering demonstrable value using AI. I’m keen to learn more about hwo you’re currently using models in Search & Recommendations
Company: I’ve heard great things about the Seek culture being super supportive. I like that Ian Narev is now CEO and am excited to see what he can bring. I also think the opportunities to expand outside teh historical core are huge.
Why did you build a startup?
I saw an opportunity to solve some of the problems that younger people in particular have with insurance - starting in the health and disability space.
Problems: affordability and trust. Using AI to minimise operating costs and changing the business model to remove misaligned incentives.
