ABN AMRO Data Expo 2024 #2

Slide 1

Slide 1 text

ABN AMRO @ Big Data Expo How to rapidly innovate with GenAI Wouter Gielen – GenAI product manager

Slide 2

Slide 2 text

GenAI radically transformed how we deliver value (what, how, who, how much) Data science - insight - models (APIs) Product management - software AI products No direct users Output is number/test Data science teams Smaller teams End-users Front/back end IT Infra Architects Design Large teams

Slide 3

Slide 3 text

We went from a small pilot to a full scale AI product team in < 18 months Fully rolled out two GenAI products < 9 months Manage a mature portfolio of 4 different AI products supporting 13k users Grew team from 5 FTE to 25 FTE on GenAI products Set up a dedicated change management program Reached over 7,500 people in internal talks on GenAI Reviewed over 500 use case submissions Won the Accenture Innovation Award

Slide 4

Slide 4 text

Great (innovation) culture = great teams = great products

Slide 5

Slide 5 text

Our use cases

Slide 6

Slide 6 text

We rolled out call summarization to all our contact officers From pilot to full roll-out in < 9 months 1000 users >10k summaries a week 8 / 10 user satisfaction 17-09-2024 | 6

Slide 7

Slide 7 text

Smart knowledge base to search answers to customer questions 17-09-2024 | 7 Find answer to question in a few clicks Automation at scale

Slide 8

Slide 8 text

Investigate why customers call and how we can improve 17-09-2024 | 8 Understand themes Deep dive For all teams

Slide 9

Slide 9 text

Released ABN AMRO GPT for daily use of GenAI 17-09-2024 | 9 10.000 users 60% weekly active users NPS > 60

Slide 10

Slide 10 text

Our playbook for rapid innovation

Slide 11

Slide 11 text

Work on big problems 17-09-2024 | 11 Work on big problems It’s a number game. Crowd source idea generation Stay connected to industry

Slide 12

Slide 12 text

Work on big problems 17-09-2024 | 12 Work on big problems Problems we like: - Mandatory - Frequent - Growing - Costly - Are in the top 3 list of our main stakeholder - End users are really annoyed about

Slide 13

Slide 13 text

Trust the iterative learning proces First, talk to the users

Slide 14

Slide 14 text

Trust the iterative learning proces Launch as fast as possible

Slide 15

Slide 15 text

Trust the iterative learning proces Master the lost art of building minimum viable products Write the specs Timebox the specs Cut the specs  What do we want to deliver as MVP  Only build stuff that contribute to the learning you seek  How fast can we deliver this?  Which specs can we remove  Any shortcuts we can take?  Don’t touch it anymore Write the specs t the specs

Slide 16

Slide 16 text

Trust the iterative learning proces Release iteratively and incrementally

Slide 17

Slide 17 text

Trust the iterative learning proces Obsess over velocity of your team to get sh*t done

Slide 18

Slide 18 text

Trust the iterative learning proces Build together with your users

Slide 19

Slide 19 text

Trust the iterative learning proces Do things that don’t scale If you can find someone with a problem that needs solving and you can solve it manually, go ahead and do that for as long as you can, and then gradually automate the bottlenecks. It would be a little frightening to be solving users' problems in a way that wasn't yet automatic, but less frightening than the far more common case of having something automatic that doesn't yet solve anyone's problems.