Secure, governed and streamlined adoption of AI is critical as business leaders chase quick wins to fast-track pilots and position themselves ahead of the curve, says Özgür Danisman, director of solutions engineering for META at Cloudflare.
Danisman will deliver an international keynote address at ITWeb Security Summit 2026 in Johannesburg on 2 and 3 June at the Sandton Convention Centre.
ITWeb Security Summit 2026
For deeper insights into modern cyber defence strategies, register for ITWeb Security Summit 2026 in Cape Town (26 May) or ITWeb Security Summit 2026 in Johannesburg (2-4 June). These annual gatherings bring together leading local and international experts to discuss the threats, technologies and strategies shaping the future of cyber security.
He says customers are eager to invest in AI, but most markets remain at the early stages of adoption.
“There are different maturity models we see in customers, starting with shadow AI, where employees use the technology and managers try to apply governance while ensuring confidential information is not shared with AI tools,” says Danisman.
Companies without controls face legal and compliance risks. One challenge is managing shadow AI, complicated by the sheer volume of accessible AI applications. Danisman explains that these tools are perforated, making it difficult for customers to distinguish between risky and permissible use.
The next maturity phase is piloting AI tools, representing a shift from awareness to experimentation.
“This requires investment in GPUs (graphics processing units),” says Danisman, adding that organisations can do this in-house or acquire services from hyperscalers.
Competitive pressure
The next phase is driven by businesses trying to gain competitive advantage and progress quickly. Business leaders are seriously considering where to test, whether to invest in GPUs or partner with hyperscalers.
Cloud service providers offer a platform for organisations to develop AI applications and experiment without exorbitant cost, adds Danisman.
“Once these applications begin to provide a return on investment, organisations can move into production and acquire services from the cloud. It’s a win-win for all stakeholders.”
Operational stage
As businesses move from experimentation to operation, they must have a solid grasp of how the technology works. It has to be correct, says Danisman – not prone to hallucination or wrong recommendations.
Agentic AI has added a layer of complexity. Danisman stresses that companies should treat agentic AI as an extra employee, requiring governance and adherence to rules.
“It can really learn, adapt and start taking initiative by itself. You need to implement approved services as a guardrail and prevent rogue agent activity."
Agentic AI requires boundaries and specified access control. Aside from its ability to learn and take initiative, it also needs protection via an “AI firewall” against external threats like jailbreaking and poisoning, adds Danisman.
He emphasises the strategic importance of partnerships with credible, experienced service providers to ensure customers can build, protect and deploy AI safely.
“AI is a massive opportunity to acquire efficiency and growth at scale. Organisations that are late in embracing AI will lose out to competition. However, adoption must be done securely, and you need a facility to deploy this technology. A cloud layer provides this,” says Danisman.
Share
