Artificial Intelligence: Send in the Analysts

Over the coming years, nearly all analysts across all sectors and industries will likely be involved in some capacity within an Artificial Intelligence (AI) or Robotics programme.

More businesses are now looking how their company can benefit from what this new technology has to offer, and it will be the analysts that will help in both the direction and strategy. One common mistake made by many organisations is viewing AI as a technology-led activity. AI is technology driven, but it should be the business that is pushing its adoption, as it is only truly effective if it is built on sound business knowledge and functional designs.

Whether a company adopts it now or later, the implementation is inevitable. Failing to do so over the next ten years will likely see those organisations lose any competitive advantage that they might currently have. Rather than seeing AI as something that will replace, I like to view it as something that will enhance or complement existing processes. Helping to free up valuable resources, by replacing tasks that are often simple, time-consuming or resource intensive with more efficient and cost-effective robotic operations. The newly found freed up time will enable key staff to refocus to do more and offer greater value within your organisation. Often competitive advantage comes from doing things differently from that of your competitor, so by having your staff focus on value-added processes, you will likely reap the rewards.

When implementing some form of Robotics programme, it should typically include two key ingredients; the first is that the programme should be business-led, as that is where the bulk of the work will be carried out and secondly that the analysts play a key strategic role from the outset. In the past, not all companies have followed this approach, as the allure of saving millions of dollars and large-scale headcount reductions, can get leaders excited, but this approach will likely end up being a costly mistake. With leaders rushing to get AI within their organisations, it is important that before engaging with expensive consultants or even investing in technology, that analysis is carried out around what technology is available on the market and how it could be used or effectively adopted within the organisation. Some of the common mistakes could be avoided by following five simple steps.

#1 Send in the analysts

Some robotics programmes are more systems led, meaning analysts may only be involved at a later stage in the process and often when the software and platform have already been selected. In my opinion, engaging with the analysts this late in the process is a big and costly mistake. Before even looking at the available technology, all of the processes across the organisation should have been identified, assessed and quantified. Nearly all analysts within any organisation, industry or sector will be able to quickly evaluate and structure the processes, identifying any potential problems and proposing viable solutions. All of which would save both time and money in the long term.

#2 Knowing what to look for

Knowing what to look for will depend on nature of your business. The most obvious places to start is with the areas that have the largest amount of resources and processes, most likely within your operational areas. Following what your people do, by going back to the floor and watching how things are carried out can help. Utilising your subject matter experts (SMEs) is a must, however by looking at the processes first hand, provides added benefits, as there might be some situations or elements that even SMEs were unaware had changed.

During your back-to-floor exercises, you will quickly be able to begin identifying potential qualifying processes. When doing this exercise, it is important that you categorise each opportunity into one of three complexity categories

• Straightforward and easy to replicate – for processes that are relatively simple and follow a sequence or series of actions. The key factors are the inputs and outputs of the process. If for example, the inputs are digital, meaning that physical data (text) is entered into a system which can be easily understood by a computer system; then you will likely be able to do something with those processes very quickly. From an implementation perspective, any opportunities identified now would probably form your basic automation stage or phase.
• Medium complexity and requires specific functionality – these processes may be straightforward, however, if an assessment or decision needs to be carried out that requires specific knowledge or experience, then it will likely take more time and effort to implement. Like the first category, inputs and outputs are crucial. If the process and decision trees are straightforward, but the process inputs are made up of scanned paper documents, then you will likely have to look into additional functionality such as Intelligent or Optical Character Recognition (I/OCR). OCR has a high extraction rate for printed text, as the text within the image will likely use a recognisable font face. However, you have also to consider other factors such any other languages and character sets on the documents your company receives. ICR has a lower success rate, as handwritten notes are harder for a system to interpret, but most organisations are seeing less written communications in this modern age. The last thing to consider within this group is the structure and design of the how the information is gathered. If it is a form, for example, you could redesign the structure to increase the I/OCR extraction rate by making it more machine friendly. Within your programme, the opportunities identified now would form your advanced or enhanced automation stage or phase.
• Challenging and requires experience – the last category, will likely reap the biggest rewards but will also probably be the most difficult to implement. You are likely at this stage to be more within the AI space of development. Within AI, you would typically have to create a series of workflows, with decision trees. Once all built, you would need to run through (as many as possible) cases or examples, recording the likely outcomes or decisions. Based on these examples, the AI system will be able to look for the correlation between the outcome and the information provided and from that, create a probability for each scenario. You can define tolerance levels within the systems, however, should say, an assessment carried out that has a probability score of 90% or more, you could program it so that the robot would complete the series of actions or steps automatically. The opportunities identified would likely form that AI/cognitive stage or phase within your programme.

Whatever process you are looking at, it is important to write down your observations, whether the process is value added or not; if not, can it be phased out, how long on average does the process take, how many people are doing it and any associated costs

#3 Scalable and flexible

Once you have your high-level analysis, this is when you should begin looking at the various technology on offer, based on the requirements and functionality needed within your organisation. It is likely that the technology that you will use going forward will be made up of a series of software or applications. Whether you purchase it all from one supplier or use a variety of vendors, it is important that whatever the approach you adopt, that you buy a scalable and flexible solution.

By minimising your upfront costs for implementation, you will be able to create a Robotics programme, that is controlled, and that will reap sufficient benefits against any costs initially outlaid. It is also important that when you are negotiating with your technology provider that they are pricing from a strategical perspective, rather than a tactical one. You may only need a small amount of technology now, but the demand will grow as the programme implements each phase. The technology needs to be priced accordingly and flexible to allow for the growth, as and when you need it.

The risk you run by buying a technology platform that is more based on your future needs up front is that you will likely be chasing bigger benefits upfront to offset the costs and could ultimately result in delays, increased costs or even failure.

#4 Benefits based on the long, not the short term

By purchasing a platform that is both scalable and flexible, would likely mean that your costs are kept to a minimum, and any benefits you realise will be done in a controlled and organised manner. When scoping the benefits, it is important to be realistic, but also to factor in aspects such as dual running. Though most Robotics solutions are robust, there is always going to be teething problems and issues to iron out inevitably. If you extend your dual running period for as long as possible, then the new technology could be continually improved, and at the same time ensuring that your customers are not affected. Remember by keeping the technology costs aligned to what you need when you need it, means that you are not under pressure to make significant savings upfront.

#5 Phased implementation

By now you know your business needs and the opportunities that exist within your organisation. By purchasing a platform that is scalable and cost-effective, means it should not adversely affect your business case and as a result means that you can provide a more stable and controlled deployment within your business. Though the approach and analysis are sound, we need to now look at the phased implementation.

• Basic Automation– this may not provide you with significant savings, but it will provide a small scale platform and a proof of concept. During this time, you should begin to hone the skills within your organisation with regards to robotics, and your employees will be able to interact and engage with the live processes. If the change is managed effectively, you will likely see staff advocating robotics going forward, even offering up suggestions for further robotics opportunities
• Advanced or Enhanced Automation – at this phase your organisation would be getting used to robotics and automated processes. Adding new functionality and capabilities to your processes would enable significantly larger savings across the business. At this stage, if a review was carried out around improving sales and the customer experiences, you should have the opportunity to free up your experienced and valuable resources to help both enhance and create opportunities to expand your business.
• Artificial Intelligence or Cognitive Reasoning – once you have workflows and processes in place, in theory, you can turn on the learning capability that exists within your AI programs. Allowing your AI to watch, observe and gather large amounts of data, will prove valuable in the long term, as the larger the data set available to the AI, the higher the probability success rate

Final Thoughts

The opportunities that exist through Robotics are not just for large corporations and can also be easily implemented within small to medium-sized businesses. Understanding your business is key, and this is where the engaging analysts at the right time, ensures a successful implementation or deployment.
Robotics or AI should not be seen as something separate and to be placed within a separate specialist area or function. Though you will likely have people that will become experienced in the implementation of automated processes, Robotics itself should be accessible and embedded across your entire organisation.

Though change can sometimes be daunting, Robotics, in the long run, will complement your existing processes and help you create differentiators within your business to build a competitive edge. For many the perspective of a robotic future is daunting though ultimately it should be seen as positive, rather than negative, as with it will come a broad range of benefits and applications that could make daily life much easier and simpler.