It's the Age of Personalized Software

It's the age of personalized software.

A few years ago, if you had a business problem to solve via software, you had to find off-the-shelf software solutions or pay a large sum for someone to develop it for you. Anything more than a marketing website, and you were looking at a significant investment.

If you wanted pre-made software off the shelf, you had to take what was available and change your workflows to fit the software.

I recently had a client that had a very specific, one-off task that wasn't cleanly solvable by any existing software without a lot of work on their end. I was able to build software around the specific requirements in less than an hour that did exactly what was needed.

A few years ago, this wouldn't have made any financial sense. Building the same application would have taken at least 15 hours of engineering time.

This application took unstructured data from several sources and created a local relational database with primary identities and associated data, merging the data from the various sources and enriching the primary identities. Less than 50k total data points as a one-time migration effort. Too much to do by hand, but too few for an enterprise-grade software investment. Some data required manual review, but the tool made it very clear what human decisions were needed, and it could be done quickly and accurately on one dashboard.

I was able to deliver it as a fully packaged, runnable executable through my client portal that the client could download and run on their own computer, do the work they needed to do, and generate the consolidated data.

This means sensitive data never traveled through a network. It remained on the client's computer in their own systems. I built the system off synthetic data that had the same shape as the real data and used a local AI coding tool to build the application, fully ensuring zero risk of data leakage.

This application was very specific, built around this one use case. There's no plan to make it more reliable or make it usable at a larger scale for a broader use case. And that's ok. It was about a $100 investment, and the whole thing can be trashed in a week.

If it was so quick and easy, could the client have done this themselves? Maybe, but knowing how to handle data appropriately across the various tools, and knowing the target architecture and how to describe the business use case in a way that will produce good output from the AI tools, is still a practiced skill. And the cost of paying someone who knows those things is likely worth it.

The software development landscape is in an interesting place right now. AI tools are good enough to make low-risk projects without me even looking at the code. When the feature set is small, and dependencies and integrations are minimized, the overall risk is small enough to verify by behavior. But the tools aren't capable of doing more complex tasks on their own. If it's software that needs to be maintained and enhanced over time, you still need a lot of human oversight.

Some tasks absolutely can be done with throwaway software, even though plenty of software still needs to be durable. The cost of trying a custom solution has never been lower.

Please reach out if there's a very specific problem you're facing that isn't obviously solvable. I'd be happy to help. And if you're looking for a more durable software solution that will be enhanced and maintained, we can still do that too.

Image Credit

Cover image: Business professional working on laptop in a bright office space, Shixart1985, Wikimedia Commons.