> Markdown version of [/videos/100631-the-state-of-agentic-ai-adel-zaalouk?t=146](https://www.wearedevelopers.com/videos/100631-the-state-of-agentic-ai-adel-zaalouk?t=146). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# The State of Agentic AI - Adel Zaalouk

Adel Zaalouk warns that running AI agents "naked" exposes your enterprise to critical vulnerabilities. Discover how to implement strict sandboxing and phantom tokens to safely scale autonomous workflows.

- **Speakers:** Adel Zaalouk
- **Event:** Coffee With Developers
- **Published:** November 10, 2026
- **Duration:** 44:23
- **URL:** https://www.wearedevelopers.com/videos/100631-the-state-of-agentic-ai-adel-zaalouk
- **Thumbnail URL:** https://customer-goifxnzhrq3d0d54.cloudflarestream.com/ac39c535d303869e2114e8a6b649164a/thumbnails/thumbnail.jpg?time=2146400ms

## Summary

The evolution of Agentic AI has moved rapidly from simple chatbots to complex, non-deterministic workflows that interact directly with enterprise systems. While the productivity gains are undeniable, the transition introduces unprecedented security risks. Adel Zaalouk, Senior Principal Program Manager at Red Hat, highlights the tension between the "move fast" mentality of AI adoption and the rigorous security requirements of enterprise environments. As agents autonomously select tools and generate code at runtime, organizations face a critical challenge: enabling developer innovation without exposing sensitive keys, inadvertently leaking data, or allowing agents to go rogue.

To address these vulnerabilities, enterprises must shift from trusting black-box hosted models to implementing robust, layered security frameworks. Zaalouk emphasizes the necessity of treating AI infrastructure similarly to cloud-native platforms, starting with a model gateway to govern access and manage token budgets. More importantly, executing agentic workflows requires strict execution boundaries. Using tools like OpenShell and Podman, developers can establish granular sandboxing that dictates exactly which URLs, binaries, or files an agent can access. By employing techniques like "phantom tokens," organizations can mask credentials so agents can utilize secrets without ever directly exposing them within the process boundary.

Beyond infrastructure, the practical applications of Agentic AI are reshaping both personal productivity and industry-specific problem-solving. From accelerating complex healthcare diagnostics through knowledge graphs to enabling generative pair-programming via shared team harnesses, the potential is vast. However, achieving this safely requires defense-in-depth, including pre-deployment red teaming and continuous drift monitoring. As Zaalouk notes, the ultimate takeaway for developers and platform engineers is to never run agents "naked." By embracing opinionated harnesses and secure sandboxing, teams can fully leverage generative AI's emergent capabilities while maintaining enterprise-grade safety.

**Keywords:** agentic AI workflows, enterprise AI security, LLM sandboxing techniques, model as a service, AI agent guardrails, non-deterministic code execution, phantom tokens for secrets, red hat AI platform, generative pair programming, healthcare knowledge graphs, AI red teaming, open source AI harnesses, MCP servers, software supply chain hardening, AI workload isolation

## Chapters

1. **Managing remote work across global time zones** (00:00) — Coordinating schedules across global time zones ensures continuous collaboration for distributed teams.
1. **Defining agentic AI and its business value** (02:26) — Agentic workflows have evolved from simple chatbots to complex applications that drive measurable productivity.
1. **Building AI platforms for enterprise developers** (05:47) — Providing secure model access and governance helps developers integrate AI without exposing corporate secrets.
1. **Managing the inner and outer loops of AI development** (10:51) — Experimenting with models and prompts transitions into deploying secure agents to production environments.
1. **Preventing rogue agents and secret exposure** (12:38) — The non-deterministic nature of agents requires strict access controls to prevent data leaks and unexpected behaviors.
1. **Implementing sandbox boundaries for agent execution** (16:20) — Masking keys and restricting binary access prevents malicious agents from compromising the underlying system.
1. **Securing AI infrastructure at the platform level** (19:52) — Defending against covert channel attacks requires isolation strategies like node anti-affinity rules.
1. **Hosting models and managing token economics** (23:04) — Self-hosting models enables better control over safety evaluations and execution policies compared to cloud APIs.
1. **Hardening operating systems and software supply chains** (27:13) — Utilizing AI to discover and patch vulnerabilities improves the security posture of Linux distributions and dependencies.
1. **Boosting personal productivity with agentic workflows** (31:20) — Automating daily tasks requires balancing execution speed with proper security configurations for non-technical users.
1. **Applying agentic AI to healthcare and data retrieval** (35:44) — Connecting disparate patient records through knowledge graphs accelerates medical diagnoses while maintaining data privacy.
1. **Collaborating on code with shared agent harnesses** (39:26) — Deploying long-lived team agents facilitates generative pair programming and pull request reviews in secure environments.

## Related Moments

- [Introduction to agentic AI engineering and code scanning](https://www.wearedevelopers.com/videos/100448-don-t-kill-my-vibes-simple-steps-to-stay-secure-when-vibe-coding) (from "Don’t kill my Vibes - Simple Steps to Stay Secure when Vibe Coding")
  - **Thumbnail URL:** https://customer-goifxnzhrq3d0d54.cloudflarestream.com/24488ef1e000cfbf47226082fd7c1ab5/thumbnails/thumbnail.jpg?time=97581ms
- [Balancing AI adoption with emerging security realities](https://www.wearedevelopers.com/videos/100427-the-things-your-ai-isn-t-telling-you) (from "The Things Your AI Isn't Telling You")
  - **Thumbnail URL:** https://customer-goifxnzhrq3d0d54.cloudflarestream.com/4a25e80f75a9d32f936557663eb7f841/thumbnails/thumbnail.jpg?time=133895ms
- [Security integration and AI skepticism in developer tooling](https://www.wearedevelopers.com/videos/1830-wearedevelopers-live-speculaitions) (from "WeAreDevelopers LIVE - SpeculAItions")
  - **Thumbnail URL:** https://customer-goifxnzhrq3d0d54.cloudflarestream.com/2cdb4226b123a771ffa9261c5605d057/thumbnails/thumbnail.jpg?time=202197ms
- [Designing agentic AI solutions for the enterprise](https://www.wearedevelopers.com/videos/1831-ai-for-enterprise-developers-dr-damir-dobric) (from "AI for Enterprise Developers - Dr. Damir Dobric")
  - **Thumbnail URL:** https://customer-goifxnzhrq3d0d54.cloudflarestream.com/e5b87103463369e16c91d95cafe45e91/thumbnails/thumbnail.jpg?time=158845ms
- [Identifying emerging security vulnerabilities in generative AI agents](https://www.wearedevelopers.com/videos/1383-the-state-of-genai-machine-learning-in-2025) (from "The State of GenAI & Machine Learning in 2025")
  - **Thumbnail URL:** https://customer-goifxnzhrq3d0d54.cloudflarestream.com/04a1e0c8f374dfdae454595bdb830ced/thumbnails/thumbnail.jpg?time=202545ms
- [Current state of security in AI applications](https://www.wearedevelopers.com/videos/1637-delay-the-ai-overlords-how-oauth-and-openfga-can-keep-your-ai-agents-from-going-rogue) (from "Delay the AI Overlords: How OAuth and OpenFGA Can Keep Your AI Agents from Going Rogue")
  - **Thumbnail URL:** https://customer-goifxnzhrq3d0d54.cloudflarestream.com/3036d9bbcdfe68d466cc9533f8a36d5e/thumbnails/thumbnail.jpg?time=122605ms

## Related Articles

- [Exploring AI: Opportunities and Risks for Developers](https://www.wearedevelopers.com/magazine/522-exploring-ai-opportunities-and-risks-for-developers)
- [What is Agentic Programming and Why Should Developers Care?](https://www.wearedevelopers.com/magazine/625-what-is-agentic-programming-and-why-should-developers-care)
- [Stephan Gillich - Bringing AI Everywhere](https://www.wearedevelopers.com/magazine/489-stephan-gillich-bringing-ai-everywhere)
- [Infer Once, Run Everywhere: Zero-Token Architecture](https://www.wearedevelopers.com/magazine/783-infer-once-run-everywhere-zero-token-architecture)

## Related Jobs

- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Staff Developer Advocate, GitHub Security Lab](https://www.wearedevelopers.com/jobs/ext/3929243-staff-developer-advocate-github-security-lab) at **GitHub**
- [Senior Product Security Architect](https://www.wearedevelopers.com/jobs/48532-senior-product-security-architect) at **Expedia**
- [Applied AI Health Data System Engineer-Senior Manager](https://www.wearedevelopers.com/jobs/ext/3735668-applied-ai-health-data-system-engineer-senior-manager) at **PwC**
- [Applied AI Health Data System Engineer-Senior Manager](https://www.wearedevelopers.com/jobs/ext/3713316-applied-ai-health-data-system-engineer-senior-manager) at **PwC**
- [Senior AI Engineer](https://www.wearedevelopers.com/jobs/ext/4500472-senior-ai-engineer) at **Redcare Pharmacy**