Workflow First, Agent Second
Why you should reach for deterministic workflows before agents. Decision criteria, architecture comparison, and the engineering instinct that saves production systems.
Every time a team builds an agent, they should first ask: would a workflow do the job? This chapter makes the comparison concrete by implementing the same task both ways and measuring the differences.
What this chapter covers
- The deterministic workflow — building a reliable pipeline for document question answering
- The bounded agent — the same task with LLM-driven decision-making
- Same task, two ways: the comparison — accuracy, latency, cost, and debuggability metrics
- Design guidance: when to choose which — decision criteria for workflows vs agents
- Planning and single-agent design — state management and bounded autonomy
- Failure modes specific to this chapter — what breaks when you choose wrong
Code companion
The working code for this chapter is in src/ch03/:
workflow.py— Deterministic workflow implementationagent.py— Bounded agent implementationstate.py— State managementcompare.py— Side-by-side comparison runner
Get the full chapter
The complete chapter text is available in the book.