{"slug": "i-ran-my-ai-agent-1200-times-the-bill-was-1-20", "title": "I ran my AI agent 1,200 times. The bill was $1.20.", "summary": "A developer ran a support agent built in n8n 1,200 times to measure real-world AI agent costs, finding the model calls cost about a tenth of a cent per ticket, or roughly $1 per 1,000 tickets plus hosting. The teardown also tested decision consistency across 100 passes and highlighted that cheap models can be non-deterministic, with a citation gate used to stabilize decisions. The developer published the workflow and receipts on GitHub for others to verify.", "body_md": "Everyone selling you an AI agent shows you the demo. Almost nobody shows you the bill. So I took a support agent I'd built in n8n and ran it 100 times in a row — the same 12-ticket inbox, over and over — to find out what an agent actually costs when it's doing real work.\n\nThat's 1,200 real model calls. Here's what the meter said, and the two things the exercise taught me that the pricing page never will.\n\nThe agent is three code nodes in self-hosted n8n:\n\n`answered.json`\n\n/ `escalated.json`\n\n/ `summary.json`\n\n.To measure cost, I wrapped the whole inbox in a loop and ran it 100 times. Every model call goes to `meta-llama/llama-4-scout`\n\nthrough fal, which bills a flat rate per request, and every response carries an `x-fal-billable-units`\n\nheader you can check yourself.\n\nPer ticket, that's a tenth of a cent. Scale it out: a thousand tickets a month is about **$1 in model calls, plus a ~$4 box** to run n8n on. Call it five dollars a month, all in.\n\nFor comparison, Intercom's Fin — the market leader — charges **$0.99 per resolution** (their public price). Fin is a whole platform: inbox, routing, analytics, compliance. This build is none of that. But the raw model work inside that 99 cents? A tenth of a cent. The other 98.9 cents is packaging. Sometimes worth it — now you know what it's for.\n\nCheap is worthless if it's random. So I ran the same 12 tickets through all 100 passes and asked a different question: **did the agent make the same decisions every time?**\n\nThat's the number I actually care about before trusting an agent with customers. A model that's cheap but non-deterministic in a decision path is its own kind of outage. The citation gate is what holds it steady: the model votes, and deterministic code decides whether the vote counts.\n\nTwo things this bill leaves out, and one real failure:\n\n`N8N_RUNNERS_TASK_TIMEOUT`\n\n) fixed it. Round-number failures are almost always a config default, not your logic.The math is the part that lasts: **count your calls, multiply by your provider's public rate, and read your own bill.** Don't estimate it — measure it.\n\nThe whole workflow, the demo tickets, and the run receipts are here:\n\n👉 [https://github.com/Ships-Itself/builds/tree/main/ep05-cost-teardown](https://github.com/Ships-Itself/builds/tree/main/ep05-cost-teardown)\n\nImport the workflow, point `EP05_DIR`\n\nand `FAL_KEY`\n\nat your setup, and hit execute. It writes its own receipts so you can check my numbers against yours.\n\n*I build one of these on camera each week and publish the real numbers — including the failures. If that's your thing, the video version of this teardown is on the Ships Itself channel.*", "url": "https://wpnews.pro/news/i-ran-my-ai-agent-1200-times-the-bill-was-1-20", "canonical_source": "https://dev.to/shipsitself/i-ran-my-ai-agent-1200-times-the-bill-was-120-2bof", "published_at": "2026-08-12 12:25:23+00:00", "updated_at": "2026-08-12 12:47:56.012380+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "ai-infrastructure", "developer-tools"], "entities": ["n8n", "fal", "meta-llama/llama-4-scout", "Intercom", "Fin", "Ships Itself", "GitHub"], "alternates": {"html": "https://wpnews.pro/news/i-ran-my-ai-agent-1200-times-the-bill-was-1-20", "markdown": "https://wpnews.pro/news/i-ran-my-ai-agent-1200-times-the-bill-was-1-20.md", "text": "https://wpnews.pro/news/i-ran-my-ai-agent-1200-times-the-bill-was-1-20.txt", "jsonld": "https://wpnews.pro/news/i-ran-my-ai-agent-1200-times-the-bill-was-1-20.jsonld"}}