Promptflow
Microsoft's open-source toolkit for building, evaluating, and deploying LLM flows.
About
Prompt flow (microsoft/promptflow on GitHub) is Microsoft's suite of development tools for the end-to-end LLM app lifecycle: visually or programmatically link LLMs, prompts, Python code, and tools into executable flows; debug and trace LLM interactions; run batch evaluations computing quality metrics over larger datasets; integrate testing into CI/CD; and deploy flows to a serving endpoint or embed them in app code. It ships as a Python SDK, a pf CLI, and a VS Code extension, with an optional cloud version inside Azure AI.
It was designed to pair with the broader Microsoft AI stack — flows can be evaluated and deployed as plugins inside Semantic Kernel orchestration. Honest caveat, and it is a big one: Microsoft's docs now recommend migrating Prompt Flow workloads to the Microsoft Agent Framework before April 20, 2027, signaling a managed wind-down. New projects should weigh that timeline; existing users get a long runway but should plan the migration.
What it can do
- Visual and code-first flow authoring linking LLMs, prompts, and Python
- Batch-run evaluation with quality and performance metrics
- LLM interaction tracing and debugging
- CI/CD integration for flow testing
- Deploy flows as endpoints or embed in app code
Setup
pip install promptflow; optional VS Code extension and Azure AI cloud version.
Prerequisites
Python 3.9+; LLM API connections (OpenAI, Azure OpenAI, etc.).
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 70; Trending score 68/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.