Don’t just code, Train your IoT device!
Intelligence is moving to the edge of the network. The smartR™ team knows how to build products with the latest AI techniques. We are here to help you integrate AI into your product. We listen to your ideas and turn them into reality. Our engagement process ensures you stay involved from start to end.

smartR

 

How can you use AI?

With all the talk and hype around AI it can be hard to know how to apply AI to your product. The first step is to decide what problem you want to solve. How do you incorporate AI into your product? You use Rule 1.





Rule #1: Let Your Imagination Run Wild

A good place to start is by making a list of new features that you want to add to your product. This is a good time to let your imagination run wild. Don’t feel constrained by what was possible yesterday, instead think about what you could achieve if software were not a limitation. AI can read text, understand speech, see objects, and control motors. AI can think ahead to plan, and run “what if” type scenarios. AI can find patterns and complex relationships. AI can schedule, plan routes, and juggle complex resourcing issues. AI can learn to adapt to changing situations such as device aging. If there is one thing that underpins all of AI, it is the power of its adaptability. AI engines can deal with a wide range of complex data, even if they haven’t seen that exact situation before. Once you have built your list of ideas, smartR will plan with you which items can be solved by AI right now.

Rule #2: New isn't Always Best

Just aim to the red spots near his eyes

Keep shooting at these spots until the eyes open, then hit quickly both eyes with your laser beam.

Rule #3: The 80/20Rule

AI is a little different from normal software development. In normal software development, the bulk of the time and effort goes into developing and testing all the features that you want to add to the software. Integration can take a long time, especially on mature codebases. With AI, the engines typically have all of the features that you want in their initial release. These engines are typically a surprisingly small amount of code. With AI engines the work goes into training and tuning these engines. Each additional piece of performance you want from the engine can grow increasingly expensive. At smartR one of the first steps we do is to plan a test set that the engines never see during their training. Whenever we finish tuning, we grade your AI engine on how well it does on this test set. It is important to set realistic goals for how well you expect the AI engine to perform on this test set. Remember the 80/20 rule applies here, otherwise you can spend more and more money getting just a bit more performance from your AI engine.

Rule #4: Choose Your Partner Carefully

A good place to start is by making a list of new features that you want to add to your product. This is a good time to let your imagination run wild. Don’t feel constrained by what was possible yesterday, instead think about what you could achieve if software were not a limitation. AI can read text, understand speech, see objects, and control motors. AI can think ahead to plan, and run “what if” type scenarios. AI can find patterns and complex relationships. AI can schedule, plan routes, and juggle complex resourcing issues. AI can learn to adapt to changing situations such as device aging. If there is one thing that underpins all of AI, it is the power of its adaptability. AI engines can deal with a wide range of complex data, even if they haven’t seen that exact situation before. Once you have built your list of ideas, smartR will plan with you which items can be solved by AI right now.