Zespan is unclaimed —

Zespan

AI agent reliability platform for engineering teams running agents in production.·zespan.com

Zespan is an AI agent reliability and observability platform for engineering and AI Ops teams. It traces every AI agent decision, tool call, and handoff in production, and provides prompt versioning, built-in LLM-as-judge evaluations, guardrails, cost optimization, and an AI ops assistant called ZespanPilot. It instruments agents with two lines of code and works across multiple LLM providers and agent frameworks.

What it's for

Debugging production agent failuresControlling AI spend / cost optimizationCatching prompt regressions before users noticeMulti-agent delegation trackingRAG pipeline retrieval vs synthesis breakdown and faithfulness scoringCost control and budget alertsEval regression testing from real production failuresCustomer support AI SLA-aware error tracking and escalation detectionEnterprise controls (SSO/SAML, audit logs, user analytics)DevOps AI pipeline anomaly detection correlated with deployments

Features 23

A/B Prompt Testing
Agent Auto-Discovery
Agent Observability
Agent Simulations
Agentic Traces with Delegation Maps
AI Anomaly Detection
AI Cost Optimization
Evaluations (12 built-in LLM-as-judge templates)
Faithfulness, Relevance, Toxicity, Groundedness Evaluation
Guardrails (custom and pre/post protections)
Model Spend Analysis
Multi-Agent Handoff Tracking
Multi-Framework Support (LangChain, CrewAI, AutoGen, OpenAI Agents SDK, Google ADK)
Per-Agent Cost Tracking
Production Monitoring, Alert Management, PagerDuty Integration
Prompt Management
Prompt Playground
Prompt Versioning
Provider Auto-Patching (OpenAI, Anthropic, Gemini, Bedrock, Mistral)
SSO / SAML
Token Usage Analytics
Tool Call Tracing
ZespanPilot AI Ops Assistant

At a glance

free tierAPI

TypeWeb app

DeploymentCloud

PlatformsWeb

Forengineering teams, AI Ops teams, AI teams

CompanyZespan · 2025

Integrations

OpenAIAnthropicGoogle GeminiAWS BedrockMistralGroqOpenRouterLiteLLMLangChainCrewAIAutoGenGoogle ADKLlamaIndexVercel AI SDKPydantic AIHaystackMicrosoft Semantic KernelOpenTelemetryPagerDutySlack

Resources

Docs

Pricing

Scale plan is custom/quote-based; exact price not published. Free tier requires no credit card.

FreeFree

No credit card required.

10,000 traces/month, 14-day retention, 2 projects, 1 seat

Solo$39/mo

100,000 traces/month, 30-day retention, Auto LLM-as-judge evaluations

100,000 traces/month, 30-day retention

Pro$149/mo

500,000 traces/month, 90-day retention, ZespanPilot, A/B testing, Custom guardrails

500,000 traces/month, 90-day retention

Team$299/mo

2,000,000 traces/month, 180-day retention, SSO, PagerDuty, 15 seats

2,000,000 traces/month, 180-day retention, 15 seats

ScaleQuote

Custom pricing, quote-based (from llms.txt pricing summary).