What it is
The Agent Development Life Cycle (ADLC) is a structured methodology for planning, building, testing, deploying, and maintaining AI agents within enterprise environments. It mirrors traditional SDLC practices but adapts them for agent-specific concerns such as soul/identity design, skill assignment, tool gating, autonomy levels, guardrail configuration, memory architecture, and continuous learning through procedural feedback loops. In Salesforce ecosystems, the ADLC often intersects with Agentforce deployment patterns, sandbox-to-production promotion workflows, and human-in-the-loop approval gates.
Why it matters
Following the ADLC ensures agents are reliable, secure, and aligned with business goals rather than ad-hoc experiments that create risk or maintenance debt. For Salesforce teams, it provides clear gates and promotion paths that integrate with existing DevOps practices, reducing the chance of agents behaving unpredictably in production.
Key components
- Agent identity and scope design
- Skill and action assignment
- Guardrail and autonomy configuration
- Sandbox-to-production promotion
- Human-in-the-loop approval gates
How it connects
In Salesforce, the ADLC maps directly onto Agentforce's configuration and deployment tooling — from defining an agent's topic scope and actions in sandbox to promoting it through change sets or DevOps Center into production with guardrails intact.
Good to know
Teams that skip formal ADLC stages often discover gaps only after an agent goes live — such as missing escalation paths or untested edge cases — so treating agent launches with the same rigor as a software release prevents costly rollbacks.
Related terms
Agentforce
Salesforce's AI agent platform that enables businesses to build, customize, and deploy autonomous AI agents across sales, service, marketing, and commerce.
Guardrails
The boundaries you set on AI agents to control what they can discuss, what actions they can take, and when they must escalate to a human.
Human-in-the-Loop (HITL)
A design pattern where AI systems require human approval or intervention at critical decision points before taking action.