Software has escaped the database
For most of my career, software revolved around databases. Data had to be collected, cleaned, structured and put into tables.
It needed a field name, a format and rules.
Is it a number or is it text?
How many characters? Should it have character??
What happens if someone enters letters instead of numbers?
....and if the data didn't fit the structure, the system didn' t work.
AI is (has) fundamentally changing that.
We can now start to think about data as simply one category of information.
Books, legislation, policies, procedures, pictures, PDFs, meeting minutes and decades of organisational knowledge can increasingly be used programmatically without first converting everything into predefined fields and tables.
Instead of always telling a system precisely what to find, AI allows us to ask it to interpret what the information means.
That's a profound change.
For the first time software can work with information that was previously written for people not machines.
Take a body corporate or strata scheme for instance. It has legislation, by-laws, committee minutes, contracts, policies and years of written decisions Historically, much of that information existed but wasn't readily usable by software.
Now it can be.
We can bring those sources together and ask questions such as Can the committee approve this expenditure and if so what happens next, then what are the steps for approval?
The system can consider the relevant information and help reach an outcome without every possible scenario having been coded in advance.
But there's a catch.
If AI can interpret almost anything, the quality and provenance of the information become more important, not less.
An AI system drawing from legislation, approved policies and trusted organisational information is very different from one drawing indiscriminately from the internet. Opinions can become facts. Old information can become current. Errors can be repeated with extraordinary confidence.
That's why the next phase of AI isn't just about smarter models.
It's about trusted information, guardrails and goodsystem architecture that allows for checks and balances.
Data is not going anywhere.
However, it's become part of something much bigger.
For the first time, software has escaped the database.
David Scognamiglio is the founder of PolicyBox. After more than 25 years working in data, digital identity and decisioning, he is exploring how AI is changing the way organisations use information to make trusted decisions.