Real AI vs. Automation in Aviation Software

Every software company is now claiming to do “AI”. The fact is that there are not many areas where AI really plays in the highly regulated (and necessarily so) aerospace industry. Automation on the other hand is a helpful tool. Where can AI really help? First we should explore the difference between AI and automation. Automation is the ability to perform task automatically. For example: If I receive a component automatically send an email to the sender letting them it’s been received and then alert the shop that a component has arrived. That is automation and it’s quite helpful in reducing the workload on people that need to perform that task. However, it is built into the software with a set of specific rules that say, “If this happens; Then make this happen…”. It does not adapt to changing circumstances nor learn new behaviors when required. It does not “learn” that now you have a separate requirement and are also checking to see if you have the applicable manuals to be able to work on that part and that they are current. “AI” on the other hand is a system that can mimic the way people reason. It is capable of learning and adapting. It does this by looking at data and if it sees that nearly every time that a component is received in inoperable condition you also document the latest revision of the manuals and check if they are current, it can learn to suggest or do that for you. It does this by looking at data and adapting the process according to the findings. How does this apply to AIQuote? AIQuote does automation like many of our competitors. It can capture a “Request For Quote” (RFQ) sent from a listing service and automate a response being sent back. But whereas our competitors do this for a limited number of components for which you have defined a specific price for, AIQuote goes several steps further. We won’t go into the specifics of our AI system (we don’t want to expose our secret sauce) but suffice it to say that AIQuote looks at all available data to come up with the best pricing strategy to win the business at the most profitable point. It suggests a price to quote and gives you a reliability-score for that price based on the amount, type and freshness of data it found. It can then perform automation in sending the quote back automatically (if the reliability is high enough) and bringing the analyst (or “salesperson”) the data to verify if the price is correct. It then learns from changed suggested prices to perform a better analysis next time.