AI = TASK DEATHMATCH.
On the right, we have tasks that a dealer (or shopper) needs to get done
On the left, we have Artificial Intelligence.
Each of our boxers has a trainer/manager (i.e. an expert in their domain). The dealer manager has a highly complex operating system that hard to penetrate. The dealer manager has been at this job for decades and the AI manager knows nothing of the Dealer Manager's daily chaos.
The AI manager goes into the belly of the dealership and discovers that a lot of the tasks are highly repetitive. The manager of AI knows his fighter is a noob and he can't take on the big big tasks the dealer and the shopper wants. The AI manager observes there is a lot of guesswork in sales and marketing, far less in fixed ops.
The AI manager gets his ear bitten by the Dealer Manager as he describes the importance of the Service department throughput and how the service writers are an insurmountable choke point to the customer experience. AI manager now makes fixed ops his top concern. Both managers work jointly to create a master list of repetitive tasks and flag those that could be automated.
The AI manager is strategic, he creates a map of tasks in fixed and variable to see those that are interconnected and those that have dependencies. The AI managers goal is to create a roadmap to conquer tasks one by one, with the intent that a basket full of automated tasks can interconnect and become a fully automated AI process (text example: "Mr Smith: NEW Service Update: your car will be completed at 4:15, 1 hour before it's scheduled completion, would you like us to send the shuttle earlier or keep the same pick up time?). #BOOM
Once the task map is tested and hardened, then the 2 managers pick a basket of tasks that'll bring the highest ROI for the dealer with the lowest degree of difficulty for the AI manager. Once work begins, the AI manager fights his way up from the bottom of the map until all tasks in the basket are up and running.
How Long before AI works its way in?
Today: ChatBots* (DI Alex, and lots of FB bots).
Coming Soon?: Fixed ops solutions (like John Quinn has proposed) have to be in development somewhere??
Coming Soon?: Fixed ops: Uber-like simplicity comes to shuttle scheduling
Coming Soon?: Variable: AI should be able to price your inventory better than you can.
Coming Soon?: Inventory, AI should be able to buy your inventory better than you can.
Summary Part 1: Look for AI to take on the small stuff (that no one wants to do) and slowly creep up the org. As AI grows and matures, look for pioneer dealers to change operations & reduce prices to take share.
Summary, Part 2: On important matters, look for AI to be personal assistants to key personnel. (i.e. Warning a GSM what is wrong and why he needs to act and the potential consequences of not acting)
AI is coming, slowly, systematically, and as it matures, it should make everyone's life less stressful.
*Most chatbots are not AI
It is super young too. Like you said on last week's Friday Refresh, people are a little carried away on how advanced they believe artificial intelligence to be. I think about all these different vendors who are promising AI tech as THE SOLUTION and can't stop wondering how they're going to have the insanely huge data set to feed that AI engine enough so that it learns. 
