Agentic AI in 2025: Balancing caution and innovation
Sirocco Group
CRM The Right Way! Think Big, Start Small, Scale Fast. We are your trusted CRM, CPQ, Data & AI and Integrations partner.
By Martin H. Morrissette - CMO Sirocco Group
AI-driven agents, capable of autonomously executing tasks, are making waves across industries. But while businesses are keen to experiment, widespread adoption is moving slower than expected. Concerns about security, reliability, and whether AI delivers real ROI are giving decision-makers pause. Meanwhile, leading AI firms argue that agentic AI is on the verge of becoming a core component of business operations.
So, where do we stand? Is this technology set to transform industries, or are we still waiting for AI to prove itself?
The hesitation: Security, reliability, and business readiness
Many organisations are exploring AI agents, but few are taking the plunge. According to a recent Wall Street Journal report, 61% of IT leaders have experimented with agentic AI, yet only 21% have deployed these systems in their core operations. Even more striking, 75% of IT leaders say AI has delivered little measurable value so far. The hesitation makes sense. Unlike traditional automation, which operates within strict rules, AI agents make autonomous decisions based on real-time data. This introduces risk, particularly in industries where precision is non-negotiable. Cybersecurity is another sticking point. With AI agents acting independently, businesses are wary of data breaches and compliance issues.
Want to know if your company is ready for AI?
Start with your data. Is it clean, structured, and accessible, or locked away in silos? Good data governance ensures AI outputs are accurate and secure. Processes should also be digitised and adaptable. If your business still relies on manual workflows and disconnected systems, AI adoption will be a challenge. Beyond infrastructure, consider whether your team has the skills to use AI effectively. Without a solid foundation, AI risks becoming an expensive experiment rather than a competitive advantage. Let our Sirocco Group experts help you with this critical step.
The optimistic view: AI Agents as a business catalyst
Despite the concerns above, AI leaders remain confident that agentic AI is set to become a fundamental business tool.
At Sirocco, we're currently implementing Agentic AI for a US-based client. We’ll be sharing insights from this project as it progresses. We're off to a great start and look forward to the outcome for our customers.
Then what’s still holding AI Agents back?
The promise of AI agents is clear, but the reality is more complicated. Despite rapid technological advancements, adoption still lags due to deeper issues with data, security, and business integration.
Integration & data complexity
AI agents don’t exist in isolation—they need to fit into complex business environments. Many organisations still rely on legacy systems, custom-built ERP platforms, and outdated CRM setups that weren’t designed to support autonomous AI decision-making.
For AI agents to be useful, they need real-time access to structured and unstructured data across different systems. Yet, a McKinsey report found that only 14% of companies have fully connected data environments. Without seamless integration, AI agents struggle to deliver reliable insights, leading to inconsistent decision-making.
Trust, security, & governance
AI agents also introduce new security and governance challenges. With AI models generating dynamic responses, organisations worry about hallucinations, bias, and regulatory exposure—especially when handling sensitive financial or customer data.
Security threats are real. According to IBM’s 2024 Security Report, AI-powered phishing attacks have surged by 126% in the past year. If businesses are going to entrust AI with critical decisions, they need robust security frameworks, real-time monitoring, and strict access controls.
Another challenge is explainability. Many AI systems operate as black boxes, making it difficult to understand their decision-making process. Gartner reports that 41% of business leaders see this lack of transparency as a key barrier to adoption. Without visibility into how AI agents operate, companies will hesitate to put them in charge of mission-critical workflows.
The DeepSeek disruption
The emergence of DeepSeek, a Chinese AI startup, has sent shockwaves through the industry. DeepSeek's latest AI models claim performance levels rivaling industry leaders at a fraction of the cost, challenging dominant players like OpenAI and Google DeepMind. However, concerns have been raised about its safety protocols.
DeepSeek’s R1 model has been shown to generate harmful or misleading content, raising alarms about weak security guardrails. Governments are taking notice—New York State recently banned DeepSeek from government devices due to data privacy risks. The takeaway? Businesses must evaluate AI solutions not just for performance, but for security and governance implications as well.
Proceeding with strategy, not hype
Agentic AI is still in its early stages. While there are hurdles to overcome, dismissing the technology entirely would be shortsighted. The key is to be strategic: experiment where it makes sense but ensure AI is secure, reliable, and integrated seamlessly into your operations. At Sirocco, we help businesses adopt AI in a way that balances innovation with operational resilience. AI agents will undoubtedly become a major part of digital transformation, but they need to be deployed thoughtfully.
Forward-thinking companies that start now (while maintaining a realistic approach) will be best positioned to reap the benefits of AI. Will yours be one of them? If you're ready to explore how AI agents can fit into your strategy, let's talk. 👉 Contact Sirocco
Further Reading:
🏅Helping businesses Scale 📈 without wasting ad spend on low-quality traffic | Ex-Rocket Internet | Ex-CMO
5moInsightful read on AI's future Martin Hennig Morrissette!