# Python Async API Use `vl_convert.asyncio` when the caller already runs an asyncio event loop, such as an asynchronous web service. Its conversion and configuration functions are awaitable, so the event loop keeps serving other tasks while VlConvert works. The synchronous and asynchronous modules share one converter and worker pool. Configure that shared state once at startup, before concurrent conversions begin. ```python import asyncio from pathlib import Path import vl_convert.asyncio as vlca spec = { "data": {"values": [{"category": "A", "value": 2}, {"category": "B", "value": 5}]}, "mark": "bar", "encoding": { "x": {"field": "category", "type": "nominal"}, "y": {"field": "value", "type": "quantitative"}, }, } async def main(): await vlca.configure(num_workers=2) await vlca.warm_up_workers() png, svg = await asyncio.gather( vlca.vegalite_to_png(spec, scale=2), vlca.vegalite_to_svg(spec), ) Path("chart.png").write_bytes(png) Path("chart.svg").write_text(svg, encoding="utf-8") asyncio.run(main()) ``` A pool runs as many conversions at once as it has workers, and extra calls wait for a free worker. See {doc}`memory-management`. See the {doc}`../async-api-reference` for which `vl_convert.asyncio` functions are awaitable and which are synchronous.