API Reference

API Reference

Complete API reference for pyai.

Quick Links

Module	Description
Configuration	Global configuration
One-Liner Functions	Simple AI operations
Agent	Agent class
Modules	Module reference

---

Configuration

configure()

Configure PyAI globally:

import pyai

pyai.configure(
    api_key="sk-...",           # API key
    provider="openai",          # "openai", "anthropic", "azure"
    model="gpt-4o-mini",        # Default model
    azure_endpoint="...",       # Azure OpenAI endpoint
    temperature=0.7,            # Default temperature
    max_tokens=2048,            # Default max tokens
)

Environment Variables

Variable	Description
OPENAI_API_KEY	OpenAI API key
ANTHROPIC_API_KEY	Anthropic API key
AZURE_OPENAI_API_KEY	Azure OpenAI API key
AZURE_OPENAI_ENDPOINT	Azure OpenAI endpoint
AZURE_OPENAI_DEPLOYMENT	Azure deployment name

---

One-Liner Functions

ask()

Ask any question:

from pyai import ask

answer = ask("What is the capital of France?")

# With options
answer = ask(
    "Explain quantum computing",
    detailed=True,
    format="bullet",
    model="gpt-4"
)

Parameters:

Parameter	Type	Default	Description
question	str	required	Question to ask
detailed	bool	False	Comprehensive answer
concise	bool	False	Brief answer
format	str	None	"bullet", "numbered", "markdown"
model	str	None	Override model

research()

Deep research on any topic:

from pyai import research

result = research("quantum computing")
print(result.summary)
print(result.key_points)
print(result.insights)

Parameters:

Parameter	Type	Default	Description
topic	str	required	Research topic
quick	bool	False	Quick summary only
depth	str	"medium"	"shallow", "medium", "deep"
focus	str	None	Focus area

summarize()

Summarize text or documents:

from pyai import summarize

summary = summarize("./document.pdf")
summary = summarize(long_text, max_words=100)

extract()

Extract structured data:

from pyai import extract

data = extract(text, fields=["name", "date", "amount"])

translate()

Translate text:

from pyai import translate

result = translate("Hello world", to="spanish")

generate()

Generate content:

from pyai import generate

content = generate("blog post about AI", style="professional")

chat()

Interactive chat:

from pyai import chat

session = chat(system="You are a helpful assistant.")
response = session.send("Hello!")

---

Agent Class

Create intelligent agents:

from pyai import Agent

agent = Agent(
    name="my-agent",
    instructions="You are a helpful assistant.",
    model="gpt-4",
    skills=[my_skill],
    memory=True
)

result = agent.run("What can you help me with?")

See Agent for full documentation.

---

Modules

Easy Module

Function	Description
ask()	Ask questions
research()	Research topics
summarize()	Summarize content
extract()	Extract data
translate()	Translate text
generate()	Generate content
chat()	Interactive chat
agent()	Create agents

Core Module

Class	Description
Agent	Agent class
LLMProvider	LLM interface
Memory	Memory management

Blueprint Module

Class	Description
Workflow	Define workflows
Pipeline	Pipeline processing
Blueprint	Complex patterns

Integrations

Module	Description
vectordb	Vector databases
sessions	Session storage
plugins	Plugin system

---

Return Types

ResearchResult

@dataclass
class ResearchResult:
    summary: str
    key_points: List[str]
    insights: List[str]
    sources: List[str]
    confidence: float

ExtractResult

@dataclass
class ExtractResult:
    fields: Dict[str, Any]
    confidence: float

---

Error Handling

from pyai.errors import (
    PyAIError,
    ConfigurationError,
    APIError,
    RateLimitError,
    ValidationError
)

try:
    result = ask("...")
except RateLimitError as e:
    print(f"Rate limit: retry after {e.retry_after}s")
except APIError as e:
    print(f"API error: {e.message}")

---

Full Documentation

For complete API documentation, see:

• Quick-Start - Getting started guide
• Configuration - Detailed configuration
• Agent - Agent class reference
• Workflows - Workflow patterns