AI infrastructure through one unified API
One configurable infrastructure layer for running AI across cloud, private cloud, on-premise, and on-device environments.
Your first request
Install the SDK
pip install backboard-sdk # or: npm install backboard-sdkSend a message
from backboard import BackboardClient
client = BackboardClient(api_key="YOUR_API_KEY")
response = await client.send_message(
"What is the capital of France?",
llm_provider="openrouter",
model_name="moonshotai/kimi-k2.6",
stream=False,
)A thread and assistant are created for you. The full walkthrough is in the quickstart.
Build AI without rebuilding the stack
The pieces teams usually stitch together by hand, behind one key.
Backboard
Single unified API
DIY stack
You glue it all together
Model routing
17,000+ LLMs
Multi-provider
LiteLLM / OpenRouter
Self-managed
State
Persistent state management
Zero-config
Redis / Postgres
DIY schema
Memory
Native memory (Lite & Pro)
Auto-extract
Mem0 / Zep / custom
Complex setup
RAG
RAG + document processing
Hybrid search
Pinecone + LlamaIndex
Manual tuning
Embeddings
Embeddings built-in
Swap-free
OpenAI / Cohere Embeddings
Separate billing
Tool calling
Tool calling & parallel execution
Native
LangChain tool wrappers
Glue code
Multi-agent
Multi-agent + portable memory
Cross-agent
Custom agent orchestration
Build yourself
Context
Adaptive context management
Automatic
Manual context windowing
Roll your own
Start free, scale with usage.
Free
$0 USD / month. API, R-CLI and the web and mobile workspace, basic memory and RAG. No credit card required.
Education
$5. Lower-cost access for students and researchers, memory and RAG, Active Context Management.
Developer
$20. API, R-CLI, desktop, and web and mobile access, advanced memory and context tools, BYOK support.
Scale
$400. Team workspace, higher usage capacity, shared billing, priority support.
New to Backboard?
Pick Studio or the Stateful API, then create one account for both.