Outshift is the incubation engine at Cisco, driving innovation in emerging technologies such as the Internet of Agents, Quantum, and Next-generation infrastructure. The Platform Engineering team at Outshift offers foundational platform services to accelerate various incubation projects.
Platform Engineers manage complex, distributed cloud-native SaaS environments involving multiple heterogeneous systems. Monitoring and diagnosing issues in these systems often requires rapidly locating information across these runtime environments, telemetry systems, and documentation sites.
The small and mighty team of Platform engineers at Outshift had to context-switch and service frequent developer requests, ranging from access management to infrastructure provisioning, while developing new features to advance the platform. This led to:
- Long wait times for request fulfillment of simple and frequent requests, often taking days to complete.
- Increased cognitive load due to constant context switching between tools and workflows.
- Operational inefficiencies, where high-value engineering tasks were deprioritized in favor of routine platform maintenance.
JARVIS: The AI Platform Engineer
To boost productivity by 10x, the Cisco Outshift Platform Engineering team developed JARVIS AI Platform Engineer, leveraging LangGraph to build a hierarchical supervisor multi-agent system. Additionally, they integrated LangSmith for tracing and agentevals for evaluation and benchmarking.
Key Features of JARVIS
- Knowledge Management: JARVIS integrates with knowledge bases (docs, policies, Jira, code, etc.) using traditional RAG and GraphRAG to quickly extract insights from unstructured and structured platform data.Z
- Self-Service Capabilities: With multi-agent LangGraph, JARVIS automates many commonly requested developer tasks such as CI/CD requests, Resource provisioning, etc.
- Code Generation: JARVIS simplifies Kubernetes deployments by generating configurations through a hybrid ML approach, translating natural language inputs into K8s manifests.
AI-Driven Developer Experience @ Outshift
Outshift integrated JARVIS across multiple developer interfaces to maximize accessibility:
- Jira: Assign a Jira task to JARVIS AI Platform Engineer and autonomously solve tasks and reaching out to submitter for additional input.
- Backstage: A chat assistant within our developer portal for seamless workflow execution.
- Webex: A secure interface for notifications and instant messaging interactions.
- CLI: Developers interact with JARVIS via the command line to build and deploy sandbox applications.
Impact of Jarvis at Outshift
Jarvis has revolutionized Platform Engineering at Outshift. Their team has experienced productivity gains as follows:
- Tasks that previously took a week, such as setting up CI/CD pipelines, can now be completed in under an hour.
- Provisioning resources (e.g., S3 buckets, EC2 instances, LLM access keys) is now instantaneous, reducing what used to be half-day tasks to just seconds.
- The need for back-and-forth communication between developers and Platform Engineering for mundane tasks has been eliminated, thanks to Jarvis's ability to guide developers autonomously or search for information.
- The organization can now manage a significantly higher volume of requests with the same team, while also reducing burnout.
Key Learnings in Agentic AI
- Internet of agents and Multi-Agent Systems unlock true potential: The future is in the multi-agentic internet of agents where the integration of first-party and third-party agents facilitates the automation of complex Platform Engineering workflows.
- Seamless UX Integration is essential: Integrating seamless user experience is essential for agentic workflows, as it will drive adoption by combining GenAI output with traditional user interfaces and task management systems.
- Reliable agentic systems with continuous evaluation and benchmark: Deploying reliable agentic systems requires thorough and continuous evaluation and benchmarking agentic systems to achieve highly accurate results.
The Future of Agentic AI in Platform Engineering
Cisco Outshift is at the forefront of integrating agentic AI into platform engineering, building an ecosystem where AI agents amplify human potential, enhance collaboration, and accelerate innovation. Their work with JARVIS is just the beginning— stay tuned as the Cisco Outshift team continues to push the boundaries of AI-powered platform engineering.
Interested in learning more?