Skip to main content
The Agent Development Lifecycle (ADLC) provides a structured, end-to-end approach to building, evaluating, deploying, and governing agentic applications on the Backbase Agentic Platform.

What is ADLC?

ADLC is a comprehensive lifecycle management framework that guides you through four sequential phases, with continuous monitoring and governance applied throughout:

ADLC Phases Overview

PhaseKey ActivitiesTools & Technologies
PrepareData ingestion, ML feature preparation, API unificationData pipelines, MCP servers, API management
BuildFramework selection, agent implementation, tool integration, RAG setupAgno, LangChain, LangGraph, MCP, Vector DBs
EvaluateUnit tests, LLM judges, human annotations, tracing analysisLangfuse, LangWatch, Promptfoo, Red teaming
RunCI/CD deployment, API exposure, monitoring, scalingArgo CD, APIM, Kubernetes, Observability stack
Continuous Monitoring and GovernanceGuardrails, security scans, compliance checks, performance monitoringAI Gateway, Nemo Guardrails, Observability tools, Security tooling

Benefits

Structured Approach

Clear phases reduce complexity and ensure nothing is missed.

Quality Assurance

Evaluation, monitoring, and governance at every step.

Production Ready

End-to-end tooling ensures agents are ready for production.

Getting Started

1

Understand the Phases

Review the ADLC Phases documentation to understand what happens in each phase.
2

Start with Prepare

Begin by preparing your APIs and data. Ensure MCP servers are available and data pipelines are set up.
3

Use Starter Kits

Leverage Starter Kits to jumpstart the Build phase with pre-configured templates.
4

Follow CI/CD

Use the platform’s CI/CD Workflows to automate the Run phase.
Best Practice: Start with a simple agent (Level 0-1) to understand the lifecycle, then progress to more complex implementations.