Contd.ai Documentation
Build durable AI workflows that never lose their way. Contd.ai provides the deterministic "Save State" engine that prevents Context Rot and ensures your agents survive failures.
Quickstart Guide
Get up and running in 5 minutes with your first workflow
Architecture
Understand how Contd.ai provides durable execution
API Reference
Complete REST API, gRPC, and SDK documentation
What is Contd.ai?
Contd.ai is a Continuity Layer for Enterprise AI that solves the fundamental problem of long-running AI agents: Context Rot.
As AI agents execute complex, multi-step workflows, they accumulate context that degrades their reasoning ability. Contd.ai separates Intelligence from Execution, ensuring your LLM stays "fresh" at step 100 because we only feed it the exact snapshot it needs.
Key Features
Durable Execution
Workflows survive infrastructure failures, API timeouts, and server restarts. Never lose progress again.
Serialized State
Work is treated as typed objects in a rigid state machine, not messy chat history.
Deterministic Resumption
If a process dies, Contd.ai reloads the last valid checkpoint instantly.
Full Auditability
Every decision is backed by an immutable, append-only ledger for compliance.
Quick Example
from contd import Client, workflow, step
client = Client(api_key="your-api-key")
@workflow
async def ai_support_agent(ctx):
# Each step is checkpointed automatically
intent = await ctx.run(classify_intent, ctx.input["message"])
context = await ctx.run(retrieve_context, intent)
response = await ctx.run(generate_response, context)
return await ctx.run(validate_output, response)
# Register and trigger
client.register(ai_support_agent)
run = await client.trigger("ai_support_agent", {"message": "Help me!"})
# If this crashes, it resumes from the last completed step
print(f"Run ID: {run.id}, Status: {run.status}")