Parallel Workflow - Definable AI

Run multiple steps at the same time. All steps receive the same input.

from definable.agent import Agent
from definable.agent.workflow import Workflow, Parallel, Step

tech_analyst = Agent(model="gpt-4o", instructions="Analyze technical feasibility.")
biz_analyst = Agent(model="gpt-4o", instructions="Analyze business viability.")
legal_analyst = Agent(model="gpt-4o", instructions="Analyze legal and compliance risks.")

workflow = Workflow(
    name="multi-analysis",
    steps=[
        Parallel(name="analysis", steps=[
            Step(name="technical", agent=tech_analyst),
            Step(name="business", agent=biz_analyst),
            Step(name="legal", agent=legal_analyst),
        ]),
    ],
)

result = await workflow.arun("Evaluate launching an AI-powered hiring tool")

How It Works

All three analysts receive the same input.
They execute concurrently (asyncio tasks).
Results are combined as [step_name]: content in the output.

Limiting Concurrency

Parallel(
    name="analysis",
    steps=[...],
    max_concurrency=2,  # Only 2 steps run at a time
)

Parallel + Sequential

Combine parallel and sequential steps:

workflow = Workflow(
    name="pipeline",
    steps=[
        # Step 1: Research in parallel
        Parallel(name="research", steps=[
            Step(name="web", agent=web_researcher),
            Step(name="papers", agent=paper_reader),
        ]),
        # Step 2: Synthesize (receives both parallel outputs)
        Step(name="synthesizer", agent=synthesizer),
    ],
)