Now live — free to use

Know exactly what
your AI agent
is doing wrong

Log every conversation, extract user intent automatically, detect failures before users churn, and get AI-suggested prompt improvements.

See Live Demo → ★ GitHub
1 Line to integrate
$0 To get started
<2s Analysis speed
100% Open source
agentlens.vercel.app/dashboard
Total Conversations
2,847
Analyzed
2,847
Failures
312
Failure Rate
11%
Top Intents
get_refund
reset_password
report_issue
check_price
cancel_subscription
Sentiment Breakdown
positive · 18%
neutral · 71%
negative · 11%

Everything you need to
understand your AI agent

Stop guessing. Start knowing exactly what's working and what's broken.

📡
Conversation Logging

One line of Python to log every conversation. Works with any LLM, any framework — OpenAI, Anthropic, Groq, LangChain, everything.

🧠
Intent Extraction

AI automatically labels what every user was trying to do. Discover your top use cases without reading thousands of conversations.

💥
Failure Detection

Automatically identifies where your agent failed, gave wrong info, or frustrated users — before they churn.

😤
Sentiment Analysis

Know which conversations left users happy, neutral, or frustrated. Spot patterns before they become complaints.

Prompt Lab

Paste your system prompt. AI analyzes your failure patterns and suggests exactly what to change and why.

🔑
Simple API

REST API with project-scoped API keys. Works as a decorator, async client, or direct HTTP call. Zero friction integration.

Up and running in 5 minutes

No complex setup. No SDKs to learn. Just log, analyze, improve.

01 — LOG
Integrate in one line

Add our Python SDK to your agent. Every conversation is automatically captured with session tracking and metadata.

from agentlens import AgentLens

lens = AgentLens(api_key="ak_...")

@lens.watch
def my_agent(msg): ...
02 — ANALYZE
AI extracts the signal

Click "Run Analysis" or set auto-schedule. AI reads every conversation and extracts intent, sentiment, and failure reasons in seconds.

POST /analyze

→ intent: "get_refund"
→ sentiment: "negative"
→ is_failure: true
→ reason: "agent deflected"
03 — IMPROVE
Fix what's broken

See your failure patterns on the dashboard. Use Prompt Lab to get AI-generated improvements to your system prompt based on real failures.

Top failures this week:
• "Can't process refunds"
• "Out of scope requests"
• "Missing product info"

→ Suggested prompt fix ✓

Start free, scale as you grow

No credit card required. Free forever for small projects.

Free
$0
forever
  • 10,000 conversations/month
  • AI intent + sentiment analysis
  • Failure detection
  • Prompt Lab
  • 1 project
  • Community support
Get started free

Ready to understand
your AI agent?

Join the waitlist. Get early access + Pro free for 3 months.

No spam. Unsubscribe anytime.
✓ You're on the list! We'll be in touch soon.
Try Live Demo →