🔗 Quick Links
- Live Tool: v0-open-claw-session-analyzer.vercel.app
- Source Code: github.com/arkbuilder/open-claw-session-analyzer
- Competition: Spring into AI Competition
🎯 The Premise
"I got a text from one of the competitors in the Spring AI Challenge, and I am not about to go down without a fight." A competitive follow-up to Episode 1, demonstrating the OpenClaw Session Analyzer built 100% by AI using v0 — and why The Effective Executive matters for managing agentic systems.
The Tool: OpenClaw Session Analyzer
A complete v0-built React visualization tool for analyzing OpenClaw session data entirely client-side. Features include:
Key Features Demonstrated
- Multi-session support: Load and compare multiple OpenClaw sessions simultaneously
- Configurable pricing: Adjust token costs per model based on your actual API usage
- Model breakdown: See usage patterns across Kimi, Auto, and other models
- Cumulative token tracking: Watch the "exponential" growth over time (concerning!)
- Highlight detection: Automatic identification of problematic spikes like that 240k token message
- Timeline view: Message-by-message breakdown with interaction highlights
- Privacy-first: Everything processes locally in the browser — no data sent to servers
The Competition Context
This video was recorded as a competitive response. After seeing another competitor's submission, I doubled down on demonstrating the OpenClaw Session Analyzer as my Week 1 entry — and hinted at more to come.
🔥 The Competitive Spirit
"I'm not about to go down without a competitive fight here. This is going to be an interesting couple of weeks." The Spring into AI competition is bringing out the best in participants — shipping fast, showing work, building in public.
Book Referenced: The Effective Executive
Peter Drucker — Two key insights applied to agentic AI management:
1. Record where the time goes: The first step toward effectiveness is procedural — knowing where you're spending time. The session analyzer does exactly this for AI usage.
2. Focus on contribution, not efficiency: "The next step... advances from the procedural to the conceptual, from mechanics to analysis and from efficiency to concern with results."
The reframe: Don't ask "how can I use AI more efficiently?" Ask "how can I be a useful contributor?" — whether to an organization, family, or project. This shifts aggravations into purpose.
What Makes This Different
Unlike toy demos that show AI doing flashy things, this tool solves a real problem:understanding where the money and tokens are going. When your average tokens per message is "way too high" and the cumulative curve looks exponential, you need visibility before you need optimization.
The guardrails help (budget limits), but the real value is the transparency —seeing that 240k token spike, understanding it was a compaction problem, and fixing it.
📝 Transcript Excerpt
"The average number of tokens per message, way too high, way too high. I have some optimization work ahead of me, but right now it just works so good, I can't help but use it more. And you can see the cumulative tokens over time continues to go, and I'm concerned that this is actually going to be an exponent."
Watch on YouTube
Episode 2 of Advisory Hour — shorter, punchier, and ready for the competition. Like and subscribe on YouTube to follow the Spring into AI journey.