Unlock the secrets of integrating AI in enterprise organisations with Jayesh Govindarajan, the SVP AI at Salesforce. Many organizations are challenged how to fit genAI into their business. Salesforce is helping it's customers by offering many integrated genAI features, but it also gives it's customers the power to build their own AI features.

Govindarjan takes us behind the scenes of Salesforce's AI journey, revealing the complex innovation of embedding AI features that elevate the user experience. We talk about AI deployment, the pivotal role of contextual data, and why developers are turning into prompt engineers to exploit the vast capabilities of large language models. We also discuss the role of the CoPilot chatbot, which is something that many users don't favor. How does Salesforce expect chatbots to become more popular?

Salesforce created many AI-ready features on its platform. Copilot plays a significant role, but there are also many AI-specific features that can be activated by a click of a button. For example, an auto-generated reply on a support ticket. Users on the platform also have the option to use Copilot at any given moment.

Organizations that want to use their own models or build their own AI features can use the new Einstein1 Studio. Salesforce also launched the prompt builder, which helps users build effective prompts in many ways, which can be really hard. According to Govindarjan, they have made a lot of progress in making prompt engineering a lot easier.

Salesforce is one of the companies that already made huge steps in AI, but it believes it's only at step one or two of ten step journey.

For this new episode of Techzine Talks on Tour we sat down with Alex Stamos during RSA Conference. Stamos is Chief Trust Officer at SentinelOne, and - among other things - also a Stanford professor and former Chief Security Officer of Facebook. We took a deep dive with him into the concept of trust.

Trust is a key component of our digital lives. It plays a role in our personal as well as in our business lives. Organizations need to know who they can trust before entering into a relationship with a vendor. At least, that's how it should be. In practice, though, a lot of time is wasted on things like vendor risk management.

The existence of vendor risk management and its accompanying forms in itself isn't a bad thing, according to Stamos. However, thinking that potential problems with some kind of product or system that is based upon tens or hundreds of millions of lines of code are going to be solved by someone filling out some kind of spreadsheet is far from realistic.

Know who to trust

One of the key questions we discuss during our conversation is how organizations can know who to trust. That's a big question that deserves a substantial discussion. We go into the role a big player like Microsoft plays in this aspect. Should you let the company that builds Windows be responsible for your cybersecurity as well? And how can a cybersecurity company prove to their customers that they're worthy of their trust? In other words, how do you provide enough transparency?

Another topic we discuss during this episode of Techzine Talks on Tour is how AI will impact the cybersecurity industry. Not only from a trust perspective, but also from an architectural and process perspective. Stamos is of the opinion that the world moves towards automated cybersecurity, thanks in large part to AI. This will give the defenders an advantage over the attackers, at least for a while.

Automation will be crucial, because of the sheer amount of defending organizations need to do. This also implies, according to Stamos, that the industry as a whole will have to move towards a centrally orchestrated way of gathering and analyzing data. Only then will it be possible for organizations to properly defend themselves. There's obviously a self-serving component to this statement, but we think fundamentally Stamos isn't wrong. It's a good concept to discuss, that's for sure.

Listen to this new episode of Techzine Talks on Tour now!

The human is often seen as the weak link in cybersecurity. You can have all the security tooling in the world, but if attackers target people they can still get in. People in general, so also employees of organizations, need to be made aware of how to spot these types of attacks. In other words, we need Human Detection and Response, i.e. HDR just as much as we need MDR and XDR.

KnowBe4 has been founded to address the challenges organizations have with educating their workforce on cybersecurity. During RSA Conference 2024, we sat down with founder and CEO Stu Sjouwerman.

We discuss the current state of affairs in security awareness inside organizations. A central theme for KnowBe4 nowadays is something they call adaptive human risk management. This is the foundation for a risk-based approach to cybersecurity through a human lens. You could see this as an extension of the many examples of risk-based cybersecurity approaches you can find in other areas of cybersecurity. The idea is to create a healthy security culture, alongside a good security stack.

There's more to it than security awareness training

We don't really discuss 'traditional' security awareness training all that much during our conversation. That's more or less table stakes now, even though there's still room for improvement on that front. However, KnowBe4 is also looking into new areas. Its acquisition of Egress is a clear example of this. Egress is a company that focuses on e-mail security. There already was a tight integration between the two players, which will become even tighter after the acquisition.

The fact that KnowBe4 ventures into the area of e-mail security may seem a bit strange at first sight. After all, KnowBe4 never was a 'traditional' security technology company, right? But if you factor into the equation that e-mail security is still the number one attack vector and that there a very strong human element to it, it actually makes perfect sense.

At the end of our conversation, we also touch on the role AI plays and is going to play in cybersecurity from a security awareness perspective. We zoom in on the rise of deepfakes. These keep getting better and better, so are harder and harder to detect for humans at the other end of them. Sjouwerman recognizes this is going to be a serious challenge, but he's also rather optimistic that common sense (and some technology of course) will eventually win this battle too.

We hope you enjoy this new episode of Techzine Talks on Tour.