Self-driving customer relationships: What marketers could learn from Tesla’s Autopilot AI

AI is becoming the new user interface.

Instead of website searches, there are personalized conversations. Rather than being placed frustratingly on hold for customer support, there is immediate help on a person’s preferred channel. Instead of mass marketing, there are individual, 1:1 relationships.

As conversational AI becomes more central to businesses’ success (Gartner predicts that by 2021, more than 50% of enterprises will spend more on bots than traditional mobile), brands need to start thinking about training their AI just like training their employees. Developments in machine learning–including deep reinforcement learning–promise to revolutionize the way businesses think about AI, making it more intelligent and human-like as compared to simpleton bots, but as the technology continues to advance, so will consumer expectations for AI-driven experiences.

The AI Building Blocks

Context, short-term and long-term “memory”

The AI should not require a person to reintroduce themselves, but rather provide an experience that gets more personal and relevant over time. The little tidbits a person reveals, whether it’s style preferences, needs / goals or even budget, should help to shape future interactions. A person should never have to tell the AI the same thing twice, unless, of course, the data or personal preference is subject to change over time.

Ongoing learning

A core part of an AI’s “DNA” is that it should continuously improve. AI learning falls info a few core categories: improved understanding of natural language and intent mapping; how to best engage with specific audience clusters in a given moment and context; and a user’s propensity to convert based on a confluence of factors. An AI’s learning is never done.

Communication with existing business systems

An AI experience can’t exist in a vacuum. By nature, it’s an incredibly personal interaction and need to tap into a brand’s CRM, commerce and other systems in order to make the experience as personal and relevant as possible. The AI should also feed the unique insights gleaned from 1-on-1 conversations back into these systems in order to create the most personal cross-channel experience.   The AI should also incorporate as many parts of the customer journey as possible. For an airline, it shouldn’t just provide travel inspiration tools, but also end-to-end booking and day-of travel support, such as rebooking and check-in. Similarly, retailers should provide guided selling and personal shopping, checkout, order tracking and support all within a single AI touchpoint. AIs true benefit comes to life when it’s powering the complete customer journey.

Human escalation protocols

As advanced as AI gets, there will always be things that a human can do better, such as complex problem solving or showing empathy in unique situations. When a situation arises in which the AI doesn’t have the skillset or emotional maturity to manage, the user can quickly get frustrated and conversations need to be seamlessly escalated to a human to take over. This handoff should happen within the same channel to cause no disruption to the consumer.

Brand safety controls

Brands need to protect themselves and make sure that they are not at risk of falling victim to merciless trolls. Therefore, every AI needs to be trained to disengage from conversations that could put them at risk. Whether it’s ending a conversation immediately when an inappropriate or sensitive topic comes up, warning users and having a three-strikes-and-you’re out policy, or elevating to a human agent, AIs need to know what topics are off limits and how to respond appropriately.

Measurements for business impact and AI performance

What good is an AI experience, if you can’t measure its effectiveness? Brands need to set goals and track conversion, whether it’s purchases, engagement, support or loyalty initiatives. Also, understanding how the AI itself is learning and optimizing over time as well as how the brand is saving operating costs typically associated with human agent support are key metrics to measure.

We’re at a critical point in the AI life cycle. Companies like Tommy Hilfiger, Target and others have shown consumers what a great AI experience can look like, and now the brands that don’t meet these raised expectations that are actually helpful risk losing consumers engaging with their AI, and their business all together.