The control plane for
agentic systems
Already have agents? Add observability in 5 minutes.
Building from scratch? Ship governed agents with the Waxell SDK.
Observe your existing agents
Add observability, cost tracking, and governance to your existing LangChain, CrewAI, or custom agents. No code changes required.
from waxell_observe import waxell_agent
@waxell_agent(agent_name="my-agent")
def run_my_agent(query: str) -> str:
# your existing agent logic — unchanged
return result
Build native agents
Build agents from scratch with durable workflows, built-in governance, and full production infrastructure.
from waxell_sdk import agent, workflow, decision
@agent(name="support-agent")
class SupportAgent:
@workflow
def handle(self, ctx):
return ctx.call(self.classify)
@decision
def classify(self, ctx):
return ctx.llm.classify(ctx.input.message,
categories=["billing", "technical"])
Everything you need to ship agents
From local development to enterprise production
Works With Any Framework
LangChain, CrewAI, or custom Python agents. Three lines of code to add full observability and governance.
Learn moreLLM Cost Tracking
Track every LLM call across every agent. See token usage, cost breakdowns, and model performance in real time.
Learn morePolicy Enforcement
Set budgets, rate limits, and approval workflows. Block, warn, or throttle agents that violate policies.
Learn moreGoverned by Design
March 2ndBuilt-in policy enforcement, approval workflows, and audit trails. Every agent action is traceable and controllable.
Preview docsDurable Workflows
March 2ndPause, resume, and checkpoint agent workflows. Built-in retry logic and failure recovery.
Preview docsProgressive Migration
Start with observe, migrate to native when ready. Every step delivers standalone value — no commitment required.
Learn moreTwo paths, one platform
Start with observability on your existing agents, or build natively with the Waxell SDK. Both paths converge on the same platform for governance, billing, and scale.
Choose your path
Add observability to existing agents or build governed agents from scratch.