3 things to consider as we embark into the GenAI journey
Organization / Economics / Technology
The advent of GenAI is most likely going to supercharge the adoption and shift from an application centric mindset to more of a “data fabric” driven enterprise landscape. In a way enterprise data already is and will become the rocketfuel for GenAI within those organizations. If they choose to do so. But they also have to be ready for it.
While this is most likely the most exciting and revolutionary period in our lifetime (from a pure technology point of view), it’ll also create entire new markets with winners and yes, lots of losers. Many new ones, but even some very well known organizations that have been around for a while may not make it. If anything GenAI will accelerate and exponentially reward or punish time-to-market delivery and failures. And the time to do this is not in the future or next year; the time is now.
Organizational Impact
New departments and teams will emerge and already do. Just as we’ve seen DevOps emerging as a new thing. It at least partially replaced IT, we now see entire departments like LegalOps, MLOps, DataOps play an even bigger role and there’ll probably be more, depending on the industry and particular theater any given enterprise is operating in.
We should also not underestimate the impact on talent and skill development. While certain jobs might go away, an entire new set of jobs and requirements will emerge. Organizations will have to adapt and change rather quickly to accommodate a whole new range of positions.
The importance of a CDO (Chief Data Officer). This role should work with the CIO, and CISO to ensure that the right data resides in the organization’s data infrastructure with proper security in place to ensure long term benefits for the organization. The CDO might even be elevated to a CAIO (Chief AI Officer).
Questions to ask:
- Who from the executive team is the DRI (directly responsible individual) in charge of enterprise data?
- Which teams are in charge and control of handling enterprise data?
- Which business and operational areas will be impacted the most by turning on GenAI?
Economics Impact
Data has been the new Oil, perhaps has been for many years. It now becomes even more valuable. However, unless you know what kind of data you actually have it’ll be difficult to place a value on it. Let alone have it ready for further consumption for GenAI.
There are several positive outcomes that can impact an organization’s bottom line by combining enterprise data with GenAI. Here are some of the big ones:
Cost reduction & time to value: It’s not just about automating everything. While GenAI does allow for greatly reducing manual data operations it can also do that rather fast and generally with a high degree of accuracy. In other words things can now get done quicker.
Increase yield and throughput: The above usually automatically leads to better yield and higher throughput. If one can do things faster, more accurately and at scale, that’s the very definition of efficiency improvements. Things to look out for here is to ensure high and consistent quality but even that, if done correctly, can be addressed in a properly setup GenAI environment.
Improved risk and compliance posture: A direct result but not to be underestimated is the reduction of operational risk and the improved compliance posture. Once you know what kind of data is ready for GenAI things like data lifecycle management, archival and retention become very important.
Questions to ask:
- Is our data classified as in, do we know what kind of data we actually have, own or should move to the cloud?
- What is our short and long term objective and strategy for using GenAI on our own data?
- What exactly do we plan to accomplish with marrying our enterprise data with a generative model?
- Which enterprise data has the potential for the best returns?
Technology Impact
Naturally, all of this plays out on the cloud. Organizations that don’t have a strong cloud strategy in place by now and at least have not made a concerted effort to get there will already face an uphill battle. Not impossible, it just requires meaningful resources e.g. funding, staffing and the right technology stack and the appropriate urgency. The main paradigm shift is coming along the lines of evaluating each point application whether it’s in a legacy on-premises environment or already in the cloud for very basic attributes like how easy it is to get data in and out of the application.
Perhaps the most pressing and immediate concern is how to make sure that proprietary enterprise data is not in any way shape or form used, intentionally or unintentionally for public models or the public domain.
The old build vs buy adage will now come up even more frequently and with more veracity both in the original M&A context but also within the individual organizations. No doubt that they will have existing tech teams but they may or may not be equipped to deliver solutions that deliver the required impact for a timely market response.
Questions to ask:
- How do we ensure that proprietary data doesn’t leak into a public model or the public domain?
- If we utilize GenAI for code generation, how do we secure our own code?
- How do we put in safeguards against model hallucinations?
- Are we as an organization cloud ready?
- Do we have the resources and wherewithal to get there?
- Is our current tech and application landscape capable of enabling GenAI?
Final thoughts
All this is playing out in real-time and with much greater impact on our future. It’s one of these situations when moving patiently and a casual wait and see attitude may not work. It could put even some of the biggest players, tech and other industries included, at a distinct disadvantage that even with all the resources in the world will be hard to overcome.
About
Martin Hack (@mhackster) is the CEO & Founder of deltazone.ai an AI powered Enterprise Data Lake for unstructured data. Before DeltaZone he was the CEO & Founder of Skytree (acquired by Infosys), a Big Data Machine Learning company. Prior to that he held executive positions at GreenBorder (acquired by Google) and Sun Microsystems.