הסוד האפל של ה-AI: למה הנתונים שלכם לא מספיקים כדי לנצח

Let's talk about the data that's the foundation for AI. Um, not in a technical sense obviously, but in a practical sense of like do you have the foundations in place that are actually going to make AI work? We ran a

piece of research recently um with a thousand Australian business leaders and 44% said access to relevant business context and data was most important for AI agents to operate efficiently. And in my experience, this is where a lot of organizations can run into trouble. There's actually a distinction that's

worth making here. Data is what happened. Eg a deal closed, a customer churned, a campaign ran. Uh context is what that data means. So why the deal closed, what made the customer leave, why your what your team learned along the way. And that's where the gap lives.

Um Angard Zero's whole AI thesis is built on being a trusted system of record. and you've talked publicly about the shift from system of record to system of action. Um what has to be true inside an organization before that shift is possible. >> Yeah, I think you mentioned obviously

data is a critical foundation. Uh but I think for any system to become the system of action uh you also have context of the workflow to your point you know there's one thing to know what is in my bank statement. It's another thing to know how many times is that the same thing? >> How often does that need to then interact with different members of my team? what does my accountant and bookkeeper do with this information? Uh they have pattern recognition. And so, you know, the next thing you need to do

on that data is trust. And to get trust, we've talked about it just in that previous conversation. It's how do you have the right human in the loop at the right time? What level do you want to be automated? And what level do you want to go in your workflow to a human to cross-check something to then trigger the next action? So we think it's critical to build that trust uh with the human in the loop and to do that I think you know we have we have

the benefit but I think any company that's in this system of record space has this opportunity which is you have years and years of context of how that information is used how it gets manipulated how it gets used to make decisions to your point um and that's all the additional context you need to build on top of the raw data >> yeah absolutely Helen can you share what

the companies that you work with have had to manage uh in terms of data foundations and what advice you would have. >> Yeah, look I think this is a journey that was already happening. Data um you know machine learning is 15 years old, data lakes and so on were already um incredibly important and the engineering of those to give access. Um I think uh

the difference now is the autonomy of the agents working with that data that potentially is fully automated. No one is checking that it makes sense. Um I I think there's a lot of talk about the data being important, the context being important. I think on top of all that is

is judgment which at the moment is more the human piece. I think it will become an autonomous piece as well and then we'll need to check the judgments >> and have checks and balances on those judgments. But at the moment it is um it is where most of most organizations are still keeping the human judgment over uh those those calls. Um I think that will change and I think that's where it gets

very dangerous if you haven't got your data um engineering strong to the point of you know it's correct you know um the logic of of the entire organization uh and can leave things autonomously making the right judgments yeah it's much more critical now >> yeah I think so so the data in the context layer the human over the top making the judgments >> at the moment. Yeah. At the moment. Exactly. Yeah.