Features Pricing Integrate Join Waitlist
CURRENTLY IN PRIVATE BETA

Know what every
AI user costs you.

Your AI bill isn't one number. It's thousands of users, each costing you differently. Tozil shows you exactly where your money goes — and how to cut it in half.

2 lines of code. Async. Never on your critical path.

tozil / cost-per-user
Users Live
7d 30d 90d
User Requests Tokens Cost Trend
sarah@acme.co 2,847 14.2M $184.30
mike@globex.io 1,203 6.8M $89.12
jess@initech.dev 956 3.1M $23.40
alex@startup.com 741 2.4M $18.90
pat@devshop.io 312 1.1M $8.20
Showing 5 of 1,247 users Total: $2,847.30 / 7d

You're flying blind on AI costs

Most AI startups look at their total bill and divide by users. That tells you nothing.

No per-user visibility

Your top 1% of users might cost 50x more than average. You wouldn't know until you're bleeding money.

Wrong models, wrong tasks

You're using Opus for tasks that Haiku handles perfectly. Every mismatched call is money wasted.

Agents go rogue

One stuck agent loop can burn through hundreds of dollars in hours. Without alerts, you find out on your invoice.

tozil / optimizations
Optimization Suggestions 3 found
Model Downgrade
save $340/mo

847 classification calls use claude-opus. Testing shows claude-haiku produces identical results for this task.

opus → haiku 99.2% quality match
Prompt Bloat
save $180/mo

System prompt in /api/analyze includes 2,400 tokens of unused tool definitions. Removing them won't affect output quality.

3,100 → 700 tokens 12 tools never called
Cache Opportunity
save $95/mo

38% of prompts in /api/summarize share identical prefixes. Enable prompt caching to reduce redundant input tokens.

1,200 daily duplicates
Estimated monthly savings $615/mo

Model Optimization

Stop overpaying
for every call

Tozil analyzes your usage patterns and finds calls where a cheaper model produces the same quality. Real suggestions, real savings.

Average customer saves 30-45% on their AI bill.

Anomaly Detection

Catch runaway
costs instantly

Get alerted the moment a user or agent starts burning through tokens. Not on your monthly invoice — in real time.

Slack, email, and webhook alerts.

tozil / alerts
Cost Anomaly Detected
ACTIVE

sarah@acme.co has spent $41.20 today — 12x their daily average of $3.40. Likely cause: agent loop in /api/research endpoint.

847 requests in 2 hours 14.2M tokens
View Details Dismiss
Cost Spike Warning 2h ago

Model gpt-4-turbo usage increased 3x since yesterday. 4 new users triggered bulk generation workflows.

Resolved 1d ago

Agent loop for mike@globex.io resolved. Total excess cost: $67.40.

You don't write code. Your AI does.

Send one link to Claude, Cursor, or any AI coding assistant. It reads the instructions and integrates Tozil for you.

01

Paste the link

Copy the integration link from your Tozil dashboard. Send it to Claude Code, Cursor, Windsurf, or any AI assistant.

// Just tell your AI:

"Integrate Tozil into my project.
 Follow: tozil.dev/integrate
 Key: tz_ak_7f3x..."
02

AI reads, AI integrates

Your AI reads tozil.dev/integrate — a page built specifically for AI agents. It adds 2 lines. Zero ambiguity.

import tozil
tozil.init()
// That's the whole integration
// Async. No proxy. No risk.
03

See your costs

Open your dashboard. Cost per user, model breakdown, optimization suggestions, and anomaly alerts — all live.

// What you see instantly:
cost_per_user$3.40
potential_savings41%
anomalies2 active

Built for AI agents

tozil.dev/integrate is a single page designed to be read by AI coding assistants. Clear instructions, no ambiguity, no multi-page docs to navigate. Your AI gets it right on the first try.

View Integration Page

Everything you need to control AI costs

Cost Per User

Know exactly what each user costs you. Track by user, team, feature, or endpoint.

Model Optimization

Automated suggestions to downgrade models where quality won't change. Real savings, not guesses.

Anomaly Alerts

Instant Slack and email alerts when costs spike. Catch runaway agents before they drain your budget.

Margin Calculator

Connect your pricing tiers. See which plans are profitable and which are losing money per user.

Multi-Provider

Anthropic, OpenAI, Google, Mistral — one dashboard for all your providers. No blind spots.

Zero Risk

Async reporting, never on your critical path. If Tozil goes down, your app doesn't notice. Uninstall cleanly anytime.

Simple pricing

Based on how much AI spend you track. Not how much you save.

Free

$0 /mo

Up to $1k tracked AI spend

  • Cost per user dashboard
  • 7-day data retention
  • 1 provider

Scale

$499 /mo

Up to $200k tracked AI spend

  • Everything in Pro
  • Margin calculator
  • Cost forecasting
  • Unlimited retention
  • Priority support

Stop guessing your AI margins

Join the waitlist. Be the first to know exactly what every user costs you — and where to cut.

No credit card. No commitment. Just early access.