ai-bridge-kit

ai-bridge-kit is a Python library to integrate AI providers through one clean API.

You can:

• Switch providers without rewriting app logic.
• Register your own local AI functions.
• Use retries and timeouts consistently.
• Keep your invention logic inside a reusable, package-ready SDK.

Install

pip install -e .

With OpenAI support:

pip install -e ".[openai]"

With Anthropic support:

pip install -e ".[anthropic]"

Install all provider extras:

pip install -e ".[all]"

For development:

pip install -e ".[dev,all]"

Quick Start

from ai_bridge_kit import AIClient

client = AIClient()

# Uses built-in local provider by default.
chat = client.chat("Explain AI integration in one line.")
print(chat.content)

emb = client.embed(["hello world"])
print(len(emb.vectors[0]))

Register your own AI functions

from ai_bridge_kit import AIClient
from ai_bridge_kit.providers import LocalFunctionProvider

provider = LocalFunctionProvider(name="my-ai")
provider.register("chat", lambda payload: "Custom answer")
provider.set_chat_function("chat")

client = AIClient()
client.register_provider(provider, set_default=True)

print(client.chat("hi").content)

OpenAI Provider (optional)

import os
from ai_bridge_kit import AIClient
from ai_bridge_kit.providers import OpenAIProvider

client = AIClient()
client.register_provider(
    OpenAIProvider(api_key=os.environ["OPENAI_API_KEY"]),
    set_default=True,
)

resp = client.chat("Give 3 startup names for an AI integration SDK.", model="gpt-4o-mini")
print(resp.content)

Additional Provider Adapters

Anthropic

import os
from ai_bridge_kit import AIClient
from ai_bridge_kit.providers import AnthropicProvider

client = AIClient()
client.register_provider(
    AnthropicProvider(api_key=os.environ["ANTHROPIC_API_KEY"]),
    set_default=True,
)
print(client.chat("Summarize agentic AI in one sentence.").content)

Ollama (local)

from ai_bridge_kit import AIClient
from ai_bridge_kit.providers import OllamaProvider

client = AIClient()
client.register_provider(
    OllamaProvider(base_url="http://localhost:11434", default_chat_model="llama3.2"),
    set_default=True,
)
print(client.chat("Explain RAG briefly.").content)

OpenAI-compatible APIs (OpenRouter/Groq/Together)

import os
from ai_bridge_kit import AIClient
from ai_bridge_kit.providers import OpenAICompatibleProvider

client = AIClient()
client.register_provider(
    OpenAICompatibleProvider.for_openrouter(api_key=os.environ["OPENROUTER_API_KEY"]),
    set_default=True,
)
print(client.chat("Give 3 names for an AI bridge SDK.").content)

CLI

ai-bridge providers
ai-bridge chat --message "What is retrieval augmented generation?"
ai-bridge embed --text "ai" --text "python"
ai-bridge call --function echo --arguments "{}"

Run tests

python -m pytest

Publish to PyPI

Use RELEASE.md for the full build + twine process.

Patent workflow support

Use PATENT_DISCLOSURE_TEMPLATE.md to document your technical novelty before filing.