𝗙𝘂𝘁𝘂𝗿𝗲 𝗽𝗿𝗼𝗼𝗳𝗶𝗻𝗴 𝗶𝗻 𝘁𝗵𝗲 𝗔𝗜 𝗲𝗿𝗮

Building on my article “Houston we have an interface problem” I want to take this 𝗮 𝘀𝘁𝗲𝗽 𝗳𝘂𝗿𝘁𝗵𝗲𝗿.

Right now we are collectively focusing on adapting our systems to work with AI, creating interfaces and bolting on AI where we can. But, I want to see more discussions around 𝗱𝗮𝘁𝗮 𝗾𝘂𝗮𝗹𝗶𝘁𝘆 𝗮𝗻𝗱 𝘀𝘆𝘀𝘁𝗲𝗺 𝗱𝗲𝘀𝗶𝗴𝗻, specifically how systems process, store and enhance your data to make it digestible for AI.

Even with the interface problem solved, AI most likely can’t make sense of the majority of our legacy corporate data. 𝗧𝗵𝗲 𝗱𝗮𝘁𝗮 𝗽𝗿𝗼𝗯𝗮𝗯𝗹𝘆 𝗹𝗮𝗰𝗸𝘀 𝗰𝗼𝗻𝘁𝗲𝘅𝘁 𝗮𝗻𝗱 𝘀𝗲𝗺𝗮𝗻𝘁𝗶𝗰 𝗱𝗲𝘀𝗰𝗿𝗶𝗽𝘁𝗶𝗼𝗻𝘀 and is probably scattered in many places and formats. This makes applying AI on it very hard. And we are still producing more of this inaccessible data every day since our systems can’t be upgraded or modified enough.

If I were to purchase or create new business software I would make sure the data is stored in as 𝘀𝗶𝗺𝗽𝗹𝗲 𝗱𝗮𝘁𝗮 𝘁𝘆𝗽𝗲𝘀 𝗮𝘀 𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝗲 𝗶𝗻 𝗰𝗼𝗺𝗺𝗼𝗻𝗹𝘆 𝘂𝘀𝗲𝗱 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲𝘀. And that the system attaches or stores relationships and meta data about the data. And that the data is both accessible via system interfaces, and exportable in simple formats that preserves the meta data.

Also, 𝗺𝗼𝘀𝘁 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗮𝗿𝗲 𝗷𝘂𝘀𝘁 𝗮 𝗰𝗼𝗹𝗹𝗲𝗰𝘁𝗶𝗼𝗻 𝗼𝗳 𝗮𝗹𝗴𝗼𝗿𝗶𝘁𝗺𝘀 or processes. Therefore I would make sure the different processes in the business systems are accessible for AI and easily composable. I’m expecting that some of the work in these processes will be replaced by AI models, 𝗻𝗼𝘄 𝗼𝗿 𝗮𝘁 𝗮 𝗹𝗮𝘁𝗲𝗿 𝗱𝗮𝘁𝗲 𝘄𝗶𝘁𝗵 𝗯𝗲𝘁𝘁𝗲𝗿 𝗺𝗼𝗱𝗲𝗹𝘀.

So, if we future proof our data and make sure our systems can be upgraded we can generate more valuable data, we can limit the risk of vendor lock in and we can adapt to new technological developments faster. 𝗧𝗵𝗲𝗻 𝗯𝘆 𝗲𝘅𝘁𝗲𝗻𝘀𝗶𝗼𝗻 𝘄𝗲 𝘄𝗶𝗹𝗹 𝗳𝘂𝘁𝘂𝗿𝗲 𝗽𝗿𝗼𝗼𝗳 𝗼𝘂𝗿 𝗲𝗻𝘁𝗶𝗿𝗲 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗶𝗻 𝘁𝗵𝗲 𝗻𝗲𝘄 𝗔𝗜 𝗲𝗿𝗮.

Thoughts or comments are more than welcome on the topic