Be the SOR, or feel the SWORD
Thesis
In the age of GenAI, be the SOR or perish.
Introduction
I was talking to a friend recently about how there may not have been the explosion in increased efficiency that was promised by Generative AI over the last few years.
I pushed back slightly -- it has fundamentally changed the way software development is completed.
We went from "isn't this a cute toy" to "what's your preferred agentic harness" to "which harness do you use in which domain" in the span of, erm, 12 months.
And now I think it has further changed -- unless your product is the system of record (SOR) at an organization, I think it's days are numbered.
What is a SOR?
A system for record (SOR) is the authorative system where data is finalized/trusted* in the organization for that kind of data. For example, your CRM would be your SOR for customer data, and your ERP (Enterprise Resource Planning) / accounting system would be system of record for finance data.
*I won't get into the whole SOR vs. SOE vs. SOT topic, but if you really want, drop us a line.
SOR is now table stakes
My thought is that the nature of what it means to be a system within an enterprise -- be it small or large, though I spend my time thinking about large enterprises with complex cloud, on-prem and generally hybrid scenarios -- has radically changed.
If you don't own a significant wedge within the organization (and not just a small wedge, which can be a feature), I think that you're not very defensible against what I call the two-devs, Claude Max, and a weekend effect. Why wouldn't they take a sprint or two to take your entire product in-house, better customized against the current state business processes, and save on long-term costs?
Why wouldn't they do this? Well, of course it's about the data. I heard this from a CIO, and it now makes sense.
A wise person once told me ...
Technology is a debt. Our data is our asset.
We haven't gone too far from "data is the new oil." Now it's "data is oil." Agentic AI is the match that can burn that oil, and your applications are the debt that might be slowing down the pace of evolution of your oil.
Aphorism overload.
Code is cheap, long live data
So if code is now cheap to create, does that mean that we don't have to worry about code? No, we still have to worry about high-quality code, but more than that, we need to worry about high-quality mechanisms to:
- be data stewards
- improve the data quality
- enhance the data quality
- enrich the data quality
- make the data quality accessible to AI
- make the data accessible to system of record
Steal this Script
Hey boss, are we the system for record for our category?If not, how can we become that?AI use disclosure
Reviewed before publishing
- Mode
- None
- Source
- my brain
- AI handled
- N/A
- Human held
- all
- Boundary
- No AI was used for this post.