AI For Managers Part 1: Hype Trainspotting


AI has been riding a hype train for the last couple of years. In this period of inflated expectations, lots of different opportunities may seem viable. The purpose of the first article in this series is to help you cut through the hype train(spotting) and help you navigate the sometimes confusing waters of AI adoption. We’re going to focus on three main points:

  1. How driving AI adoption for your organization is easier than you think

  2. Not to wait, you’re wasting your data

  3. Thinking long term

By the end of the first article you will be in a better position to assess your organization's ability to develop and implement an AI strategy. You don’t need to be a giant company like Google or IBM to create an ‘AI First’ strategy in order to benefit from the opportunities that AI can bring to your organization.

It’s easier than you think

Start Small

Starting small is a theme you will hear over and over from us. Innovation is not without risk. In order to mitigate the risk (more on this in part 2) it is important to build on success incrementally. Inevitably, there will be a variety of stakeholders: some of whom will be skeptical of the value offered by AI. Establishing value with a small win is the easiest way to build momentum inside your department or organization.

It is also important to start simple. Starting simple provides the opportunity for compounding value to be created. Choosing simple and repetitive tasks for your first experiment is critical. These tasks lend themselves to the type of AI systems that are currently available today. AI has a variety of current limitations. In order to manage these limitations, it is best to choose bite sized task chunks. These types of tasks should not have huge levels of complexity and should be conducted at a minimum operational scale of 1000’s of times per year. These processes exist in every business. Don’t believe us?


Focus on workflow intensive tasks

One of the best entry points for adopting AI in your organization is to focus on workflow or resource intensive tasks. These opportunities are often the most organic and suitable for early stage adoption. They also have the potential to yield high Return on Investment (ROI).

There is no universal solution presently available that can solve your organization's challenges in an ‘off the shelf’ fashion. AI inherently requires customization and adjustment to meet your organization's needs. Properly identifying the challenge to solve with an AI solution lets you minimize the cost of configuration and maximize the expected benefits.

We are strong believers in active learning. When we leave humans in the loop it allows us to incrementally improve the performance of the software. Human in the loop learning also allows us to build a data set without significant operational disruptions. It is for these two reasons that we always advocate for using AI to empower humans to perform at a higher level and do not contemplate projects that replicate or replace humans.

DOn’t Wait! You’re wasting data

Workflow as data

Your existing workflows are your biggest asset. Integrating information capture into your business process is critical to building a long term AI strategy and solution. Whether you are going to implement AI tomorrow or in 10 years it is important to design a data capture strategy today. Workflow capture is the most effective approach for gathering training and testing data to facilitate your AI objectives. By focusing on workflow as a capture point the generation of data is not done in isolation but rather as a component of the business process.

Data capture is another opportunity for your organization to develop compounding value. The more historical data you collect in an intelligent and organized fashion the easier the transition to AI integration will be and the lower the cost of design, development and deployment.

User interaction for annotation

Annotations are the backbone of AI. Annotations are used to teach AI algorithms. The algorithms ‘learn’ by observing many examples labelled with an ‘input’ value and an ‘output’ value. These values enable the algorithm to learn a function that is capable of mapping new inputs to new outputs.

By focusing on optimizing the front end of the data lifecycle Capture, Curate, Consume the downstream results of AI can be amplified. Better designing your capture and curate processes leads to extracting more value upon consumption. User interactions in existing workflows are one of the easiest paths to creating organizational momentum around capture and curation.

Repetition & Low level tasks

Most importantly of all for AI adoption in your organization to be a success it is important to target repetitive and low level tasks. These tasks exist in every workplace. Repetitive tasks are not tasks that humans are exceptionally gifted at performing consistently. Where these tasks involve intuition, judgment or large scale information processing is where AI can shine. AI is not needed to push a button every hour on the hour however AI is much more capable of pressing a button whenever a person gets too close to an operating piece of equipment. When human attention and focus can drift AI is more than capable of filling in the gaps.

Long term thinking

As an organization long term thinking is critical to your future success. AI is increasingly a mandatory component of long term operational thinking. This is because AI has the potential to augment your organization for the better.

AI can move humans up the value chain. By focusing on human in the loop AI we have the opportunity to augment human performance. As we shift humans to higher value activities their net productivity receives a corresponding increase. This increase in productivity means that organizations are able to operate with the same overhead costs but achieve more. What this means is that if you aren’t thinking about bringing AI into your organization your competitors are going to reap the benefits of these savings and out-compete you for your customers. Don’t believe us? Harvard Business Review agrees.

Finally, in most organizations there are tasks which would be performed, except organizations currently lack the quantity of humans to help perform them. We call these tasks ‘non-human scale’ tasks. Non-human scale tasks are any tasks that your organization cannot afford the human power to perform. For example reviewing 16,000 documents or watching 250 security cameras simultaneously or performing quality assurance checks on all calls received in your call centre.


Getting started with AI is possible for any organization. The potential for success is dramatically higher when the initial problem is at a manageable scale with easily identified ROI. Focusing on tasks which have established business processes and workflows is critical for both opportunity identification and annotation generation. Tackling the high volume, low human value tasks is one of the best places to start. The cost  of not starting today, even if that start is just developing and implementing a capture and curate process for your data, is enormous. The risk of being left behind or unprepared for the coming changes is far greater than the cost of getting started.

If you want to learn more reach out!