#003 - Services & Marketplace

June 4 2020/45 min read

We are joined by Per Klaar. He is the Chief Data Officer at Offerta, Sweden's largest service marketplace. How can AI be used to help match services? What are the key opportunities, and what are the challenges?

(The discussion is in Swedish)

Key takeaways from this episode

AI is not a field in itself. It spans everything.
  • When electricity was invented many business leaders did not know how to relate to it. It was common to appoint an electricity director. Chief AI Officer and other similar titles are the same thing. They will not exist in the future.

Problem too hard to solve. Probably a job for AI.
  • Many of the problems we had were too hard to solve with regular methods, which is why we turned to AI. Working together is important, and looking at the entire picture.
Our customers don’t see if it’s AI or not. They just get a better service.

Access to data allows for continued growth
  • A lot of real time indicators make it possible to predict the future of business. For instance, the probability to renew contracts in the future.
Focus is on AI projects that generate revenue
  • During the first two years AI was applied to internal processes; accurate predictions of volumes, and automating google advertising.
Today, many different AI applications
  • Key for business is to retain our customers
  • Classifier analyzes at what a customer is writing and suggest what category is best suited. 
  • Accurately match RFPs to contractors. Inspired by AirBnb.
  • Predict how popular particular RFPs will be when they enter the system.
  • Used customer behaviour data to get better understanding, and predict which customer we needed to cater to. And how. 
The models work fantastic right now.

Building AI is easy. Applying AI is hard.
  • Almost all time is spent on integrating models in product, infrastructure, and guaranteeing stability. Very little on modelling.
Chatbots not there yet. Good dialogue still too hard.

AI cannot be a separate department
  • Needs to be in the hands of domain experts. Distribute your AI expertise throughout your organisation, to be part of teams of other expertises. Do not keep them apart.
  • The more teams get to work with AI, the faster they become.
Hard times lead to change
  • Interest for AI powered service matching has risen when contractors see their natural inflow reduced due to COVID-19.
  • Organisations that have doubled down during this crisis will take market shares.