Transcript#
This transcript was generated automatically and may contain errors.
The biggest gap that I see is where organizations, they are mistaking this work that they're doing for actual output. So at the board level, we have all of these kind of, we're investing in AI initiatives, but when you think about the ground floor and where we're actually working, there's a lot of fragmentation still, and data scientists and analysts are working with different tools.
And until an organization is employing data science to the level where they can actually base their business decisions on, they'll always kind of be stuck in this experimentation phase. And so that's the biggest gap I think organizations need to think about.
AI productivity and agents
They're delivering really good value in terms of speeding up your productivity, being able to automatically code, being able to generate things that have a lot of that high frequency. And when you start thinking about agents, it's not so much, can an agent do this? Where it falls apart is fitting that into your actual environment.
So thinking about your governance, the different tools that you're working with, your whole environment as an organization. Can your agents actually adapt to your organization's environment? And so that's the biggest part where I feel like they fail.
Treating automation as a spectrum
The organizations that are really successful with that are treating it more like a spectrum rather than this kind of binary automate everything or just do manual processes. So they're really looking at it from two different decisions. One, do I have this process that has high frequency and low risk? Yes, let's automate this.
Do we have this other process that might be more ambiguous in terms of the result and also has a lot of risk involved in that? That's where you still want that human in the loop. And so we're not looking at everything as automate everything, throw it in there, or we're doing everything by hand. We really have to look at it from a spectrum.
The organizations that are really successful with that are treating it more like a spectrum rather than this kind of binary automate everything or just do manual processes.
Scaling data science as an organizational product
It's not seeing your data science and AI organizations as kind of like this experimentation kitchen. It's really embracing AI and data as an organizational product, embedding that into your workflows and essentially making that a repeatable process.
It might be one thing to take a model and be able to put it into production, but how easy is it for you to take more of your data science workflows and spread that across the organization and really scale? And unless those organizations are really looking at it from a holistic view, you're always going to be stuck in experimenting, prototyping, and not really getting things to that production level.