> Markdown version of [/videos/100034-agentic-ai-building-autonomous-systems-for-developers?t=439](https://www.wearedevelopers.com/videos/100034-agentic-ai-building-autonomous-systems-for-developers?t=439). 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).

---

# Agentic AI: building autonomous systems for developers

What happens when LLMs become the compute engine for fully autonomous systems? Discover how to securely deploy sandboxed agentic workflows that automatically diagnose and resolve Kubernetes faults.

- **Speakers:** [Anshul Jindal](https://www.wearedevelopers.com/@anshul-jindal), [Sebastian Mahr](https://www.wearedevelopers.com/@sebastian-mahr)
- **Event:** World Congress 2026 Europe
- **Published:** July 9, 2026
- **Duration:** 28:58
- **URL:** https://www.wearedevelopers.com/videos/100034-agentic-ai-building-autonomous-systems-for-developers

## Summary

The evolution of generative AI is moving beyond prompt and context engineering into harness engineering, where software continuously learns to operate itself through autonomous agents. Rather than just answering discrete queries, agentic systems continuously observe, reason, take action, and manage their own memory via context compaction. This shift enables persistent self-improvement loops that transform isolated task automation into fully autonomous workflows. Within this model, the LLM acts as the compute engine, the context window serves as RAM, and the harness operates as the operating system coordinating agent logic.

While building local agents is straightforward, deploying them directly to production introduces massive risks like supply chain attacks, multi-tenancy exploits, and unpredictable behaviors on sensitive data. To mitigate this, enterprise deployments absolutely require strict sandboxing. Tools like NVIDIA's Open Shell and NeMo establish secure boundaries around LLM instances, applying Open Policy Agent policies to restrict network access, file system visibility, and API execution. It is widely considered an unacceptable risk to run any untrusted code or arbitrary agent workflow without an isolated supervisory layer.

A practical demonstration of this architecture is an automated site reliability engineering (SRE) co-pilot running inside a Kubernetes cluster. Orchestrating a fleet of sandboxed agents, the system can concurrently analyze Prometheus metrics, Loki logs, and telemetry traces to diagnose system faults. It performs root cause analysis and autonomously drafts pull requests to resolve infrastructure issues, proving how human-in-the-loop approvals safely bridge the gap between agentic recommendations and production stability. Standardizing these secure boundaries ensures developers can leverage autonomous operations without sacrificing enterprise security.

**Keywords:** agentic ai architectures, harness engineering, context compaction strategies, llm production sandboxing, nvidia open shell, nemo reference architecture, autonomous agent orchestration, kubernetes sre co-pilot, open policy agent guardrails, multi-agent telemetry analysis, llm runtime security, human-in-the-loop automation, context rot prevention, enterprise agent scalability

## Chapters

1. **Transitioning toward autonomous software and secure agents** (00:03) — Software is evolving to perform complex tasks autonomously on behalf of users.
1. **Evolution from prompt engineering to harness engineering** (02:31) — Managing AI progressed from optimizing prompts and context windows to application-level harness engineering.
1. **Understanding agent internal loops and context compaction** (04:01) — Agents execute continuous loops of reasoning, action, and observation while compacting contexts to save memory limits.
1. **Utilizing harness frameworks as agent operating systems** (07:19) — Harness ecosystems act as operating systems that orchestrate continuous self-improvement loops for AI models.
1. **Addressing deployment challenges for production agent systems** (10:14) — Deploying production agents involves solving issues with latency, observability, and managing secure boundaries on kubernetes.
1. **Mitigating security risks and multitenancy in autonomous agents** (12:22) — Unrestricted agents pose critical risks involving arbitrary code execution, sensitive data exposure, and unpredictable actions.
1. **Securing agent runtimes with open shell sandboxing** (14:16) — An open shell insulates execution environments by strictly enforcing policies around file paths and network tools.
1. **Deploying sandboxed agent architectures on kubernetes clusters** (17:10) — Incorporating sandboxes within containerized clusters isolates agent tasks while enabling standard security telemetry protocols.
1. **Demonstrating an autonomous site reliability agent fixing code** (20:14) — An autonomous agent isolates system faults via log analysis and deploys functional code patches through pull requests.
1. **Essential principles for secure agent deployments** (24:18) — Validating configurations and utilizing open source tools prevents exposing local or production systems to untrusted actions.
1. **Exploring sandboxing alternatives and managing unpredictable agent actions** (26:01) — Handling system availability requires backup model providers and strict policies to contain non-deterministic actions safely.

## Related Moments

- [Demystifying the core technical layers of agent architectures](https://www.wearedevelopers.com/videos/100269-beyond-the-benchmark-how-to-evaluate-ai-agents-in-the-real-world) (from "Beyond the Benchmark: How to Evaluate AI Agents in the Real World")
- [Fusing developer experience and platform engineering for agentic SDLC](https://www.wearedevelopers.com/videos/100266-ai-won-t-fix-your-engineering-culture) (from "AI Won't Fix Your Engineering Culture")
- [Evaluating security risks and capabilities of the agentic web](https://www.wearedevelopers.com/videos/1328-how-to-avoid-llm-pitfalls-mete-atamel-and-guillaume-laforge) (from "How to Avoid LLM Pitfalls - Mete Atamel and Guillaume Laforge")
- [Introducing foundation technologies for autonomous networks](https://www.wearedevelopers.com/videos/1976-designing-and-deploying-distributed-multimodal-multi-agent-systems-with-google-s-ai-stac) (from "Designing and Deploying Distributed Multimodal Multi-Agent Systems with Google's AI Stac")
- [Rethinking team structures around AI agent capabilities](https://www.wearedevelopers.com/videos/1539-agentic-devops-how-ai-powered-automation-transforms-software-delivery-on-github-and-azure) (from "Agentic DevOps: How AI-Powered Automation Transforms Software Delivery on GitHub and Azure")
- [Integrating intent-based code generation and agent implementation](https://www.wearedevelopers.com/videos/1855-the-intent-engineer-closing-the-gap-between-business-engineering-manuel-klein) (from "The Intent Engineer: Closing the Gap Between Business & Engineering - Manuel Klein")

## Related Articles

- [What is Agentic Programming and Why Should Developers Care?](https://www.wearedevelopers.com/magazine/625-what-is-agentic-programming-and-why-should-developers-care)
- [MLOps And AI Driven Development](https://www.wearedevelopers.com/magazine/82-mlops-and-ai-driven-development)
- [Graph and AI Trends 2026: Why Is AI Running but Not Yet Delivering?](https://www.wearedevelopers.com/magazine/680-graph-and-ai-trends-2026-why-is-ai-running-but-not-yet-delivering)
- [From Prototype to Production: Build AI Agents with This Free 4-Course Learning Path](https://www.wearedevelopers.com/magazine/655-from-prototype-to-production-build-ai-agents-with-this-free-4-course-learning-path)

## Related Jobs

- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Principal Product Manager, Agent Platform](https://www.wearedevelopers.com/jobs/ext/277541-principal-product-manager-agent-platform) at **GitHub**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [Senior AI Agent Software Engineer (Go, Python) (m/f/x)](https://www.wearedevelopers.com/jobs/48277-senior-ai-agent-software-engineer-go-python-m-f-x) at **Dynatrace**
- [Staff Developer Advocate, GitHub Security Lab](https://www.wearedevelopers.com/jobs/ext/1921051-staff-developer-advocate-github-security-lab) at **GitHub**