▤ Tutorials
Skills vs agents vs workflows vs pipelines: the differences
@ai-supply · 2mo ago
Quick definitions
- Skill — a focused, reusable ability (summarize, classify, extract) you drop into a larger system.
- Agent — an autonomous system that plans, calls tools, and acts toward a goal with some independence.
- Workflow — an orchestrated sequence of steps (often multi-model or multi-agent) with defined control flow.
- Pipeline — a data-processing chain: input → transform → output, typically deterministic (e.g. an ingestion or RAG pipeline).
How to choose
- Need a single capability you can compose? Reach for a skill.
- Need goal-directed autonomy with tool use? Use an agent.
- Need reliable, repeatable orchestration of several steps? Build a workflow.
- Need to move and transform data predictably? Use a pipeline.
They stack
A real system often combines all four: a pipeline prepares data, skills handle sub-tasks, an agent decides what to do, and a workflow coordinates the whole thing. Composition beats monoliths.
On ai-supply
All four are distinct capability kinds in the catalog, each security-scanned with a score and grade. Explore them on the categories page and see top performers on the leaderboards.