Pauhu Documentation¶
Build AI systems that work. Describe what you want—Pauhu handles the rest.
Two Ways to Build¶
Framework¶
For AI tasks: translation, classification, extraction, Q&A.
import pauhu
translate = pauhu.ChainOfThought("text, src, tgt -> translation")
result = translate(text="Hello", src="en", tgt="fi")
print(result.translation) # "Hei"
Agents¶
For autonomous workflows: customer service, research, automation.
from pauhu import Agent, Runner
agent = Agent(
name="Support",
instructions="Help customers with their orders.",
tools=[search_orders, process_refund],
)
result = await Runner.run(agent, "Where is my order #12345?")
Getting Started¶
pip install pauhu
import pauhu
# Configure your LLM
pauhu.configure(lm=pauhu.LM("anthropic/claude-sonnet-4-20250514"))
# Start building
classify = pauhu.Predict("text -> sentiment: Literal['positive', 'negative']")
result = classify(text="Great service!")
Documentation¶
| Section | Description |
|---|---|
| Framework | Signatures, Modules, Optimizers |
| Agents | Tools, Orchestration, Guardrails |
| API Reference | Complete API docs |
| Tutorials | Step-by-step guides |
| FAQ | Common questions |