The Task-First Approach to AI Implementation

At ScaleUp:AI 2024, Andrew Ng shared an insight that perfectly builds on our discussion of change management from last week: AI doesn’t automate jobs – it automates tasks. This observation connects powerfully with Clayton Christensen’s Jobs To Be Done framework and helps explain why that crucial 5:1 ratio of change management to technology investment is so important.

Beyond Simple Automation

The temptation in AI implementation is to look at entire roles or departments and ask, “How can we automate this?” But this misses the crucial connection between tasks and value creation. Every job exists to serve a purpose – what Christensen calls a “job to be done” for the customer. Between that job and its outcome lies a series of tasks, each contributing differently to the value chain.

When we decided to implement AI in our support team, we focused on the first task which was answering a case. This targeted approach forced us to understand the nuances of how value is delivered through each customer interaction, rather than attempting sweeping automation. It also allowed us to focus our change management efforts precisely where they would have the most impact.

The Value Creation Chain

Consider customer service, the example Ng used in his presentation. The fundamental job to be done isn’t “answer customer queries” – it’s “help customers feel confident their problems will be solved quickly and completely.” This reframing changes how we evaluate which tasks are truly crucial for value creation.

In Ng’s analysis of customer service tasks, some had high AI potential (like handling text queries and keeping interaction records), while others showed lower potential (such as handling complex phone calls). By identifying these distinctions, we can better target both our technology investments and our change management efforts.

Finding True Productivity

This task-level analysis directly supports the 5:1 investment ratio I discussed last week. When you understand exactly which tasks you’re enhancing with AI, you can focus your change management efforts precisely where they’ll have the most impact. Instead of broadly trying to “manage change” across an entire department, you can target your efforts on specific high-value tasks and their associated workflows.

A financial services client recently demonstrated this perfectly. Instead of broadly automating customer service, they mapped each task to customer needs. They discovered that while AI could handle routine balance inquiries brilliantly, customers dealing with fraud needed human reassurance. By selectively automating tasks based on this understanding and investing heavily in change management for affected workflows, they increased both efficiency and satisfaction.

The Implementation Journey

The path forward requires a different kind of analysis. Instead of starting with the technology, start with the job to be done. Map out how different tasks contribute to fulfilling that job. Only then can you effectively allocate both your technology and change management investments.

In the customer service example, when we overlay AI potential with job importance, we find interesting patterns. Text query handling has high AI potential and directly supports the core job of quick problem resolution. But phone calls, while showing lower AI potential, often handle the complex issues where human empathy and understanding are crucial to the job to be done.

From Theory to Practice

Let me share a framework that’s emerged from working with dozens of organizations on this journey. Start by asking three questions:

First, what is the fundamental job your customers are hiring you to do? For a project management tool, it might be “help me deliver projects on time and within budget” rather than simply “track tasks.”

Second, what are the tasks that contribute to this job? Break down every interaction, every touchpoint, every process. Be granular but purposeful.

Third, how does each task either enable or inhibit your ability to fulfill that core job? This is where the magic happens – where task analysis meets customer value.

Looking Forward

This intersection of Jobs To Be Done and task-level AI analysis helps us understand exactly where to focus our change management investments. It moves us beyond simple automation to true value enhancement. The question isn’t just “Can we automate this task?” but “Will automating this task help us better serve our customers’ needs, and are we prepared to manage that change effectively?”

The real productivity gains come from understanding not just what tasks can be automated, but how that automation serves the fundamental job to be done – and then investing appropriately in the change management needed to make that automation truly invisible and effective.

Next week in Part 3, we’ll explore how this value-focused approach builds trust in AI systems by ensuring automation serves a clear purpose. Until then, I encourage you to examine your own AI initiatives through this lens: are you automating tasks that truly matter to your customers’ jobs to be done, and are you investing enough in managing that change? The answer might reshape your entire approach to AI implementation.

This post is part 2 of our ScaleUp:AI 2024 series. Read part 1 on change management and join us next week for part 3 on building trust in AI systems.