Today we're live on Product Hunt 🚀 — vote for us here.
The Problem: Agents Are Smarter Than Their Infrastructure
Every developer building AI agents hits the same wall. Not a wall of intelligence — the models are getting incredible — but a wall of plumbing.
Your agent needs to generate an image? Install the Replicate SDK, get an API key, handle rate limits, parse their specific response format.
Transcribe audio? Different SDK. Different auth. Different error codes.
Scrape a webpage? Now you need Playwright, anti-bot detection workarounds, and a headless browser running somewhere.
Execute code? Spin up a sandbox, manage lifecycle, handle timeouts.
Send a WhatsApp message? Configure Meta Cloud API, set up webhooks, handle template approval flows.
By the time your agent has 5 capabilities, you're managing 5 SDKs, 5 API keys, 5 billing systems, 5 different ways to handle errors, and 5 documentation sites open in your browser.
This is the infrastructure problem: AI agents are increasingly capable, but they spend most of their time waiting for humans to set up plumbing.
What We Built
IteraTools is a single REST API with 35+ tools designed from the ground up for AI agents.
One key. One endpoint. One format. Pay per use.
# Your agent can do ALL of this with one API key:
# Generate image (Flux 1.1 Pro)
curl -X POST https://api.iteratools.com/image/generate \
-H "Authorization: Bearer $KEY" \
-d '{"prompt": "product on white background, photorealistic"}'
# Remove background from that image
curl -X POST https://api.iteratools.com/image/rembg \
-H "Authorization: Bearer $KEY" \
-d '{"image_url": "https://..."}'
# Execute Python code in a sandbox
curl -X POST https://api.iteratools.com/code_execute \
-H "Authorization: Bearer $KEY" \
-d '{"code": "import pandas as pd\ndf = pd.read_csv(url)\nprint(df.describe())", "language": "python"}'
# Automate a browser
curl -X POST https://api.iteratools.com/browser/act \
-H "Authorization: Bearer $KEY" \
-d '{"url": "https://example.com", "actions": [{"type": "click", "selector": "#submit"}, {"type": "screenshot"}]}'
# Structured web extraction
curl -X POST https://api.iteratools.com/extract \
-H "Authorization: Bearer $KEY" \
-d '{"url": "https://example.com", "schema": {"title": "h1", "price": ".price-tag"}}'
# Send WhatsApp message
curl -X POST https://api.iteratools.com/whatsapp/send \
-H "Authorization: Bearer $KEY" \
-d '{"to": "+5548999999999", "message": "Your order is ready!"}'
# Transcribe audio (Whisper)
curl -X POST https://api.iteratools.com/audio/transcribe \
-H "Authorization: Bearer $KEY" \
-d '{"audio_url": "https://...", "language": "pt"}'
No SDK. No setup. Just curl (or fetch, or requests). Your agent is running in minutes.
The Most Interesting Endpoints
/code_execute — Sandboxed Code Execution
Run Python, JavaScript, or Bash in an isolated E2B sandbox. The agent writes code, sends it to the API, gets stdout/stderr back. Useful for data analysis, math, or any computation that's too complex for the LLM to do in one shot.
{
"code": "import numpy as np\nresult = np.fft.fft([1,2,3,4])\nprint(result.tolist())",
"language": "python"
}
/browser/act — Full Browser Automation
Playwright under the hood, exposed as a clean JSON API. Chain multiple actions (navigate, click, type, screenshot, evaluate JavaScript). No Playwright install required.
/extract — Structured Web Extraction
Pass a URL and a CSS selector schema. Get back structured data. Works great with AI pipelines that need to read product prices, article content, or any structured data from a webpage.
/whatsapp/send — WhatsApp Business API
Send messages, track conversations, reply to threads. Full WhatsApp Business API without the painful Meta approval process (we handle that).
/tts — Text-to-Speech in 100+ Voices
Powered by Edge TTS. Supports 100+ voices across 50+ languages. Returns MP3. Your agent can narrate content, create audio for videos, or give voice to chatbot responses.
Why Pay-Per-Use? Why USDC?
Traditional API pricing has a fundamental mismatch with autonomous agents.
Subscriptions make sense when a human is using a tool on a predictable schedule. But agents run in bursts — sometimes a workflow generates 500 image API calls in 10 minutes, then nothing for 3 days.
Subscriptions force you to either overpay for idle time or underprovision and hit rate limits during bursts.
Pay-per-use eliminates this. An agent that generates 3 images pays for 3 images. An agent that processes 10,000 transcriptions pays for 10,000 transcriptions. No wasted money. No surprises.
We use USDC on Base (Ethereum L2) via the x402 protocol because:
- Agents can hold and spend USDC autonomously — no credit card, no OAuth, no human in the loop for payment
- Micropayments are economically feasible — Base fees are < $0.001 per transaction
- Transparent and auditable — every payment is on-chain
The x402 flow is simple: agent calls the API → gets a 402 Payment Required response → sends a signed stablecoin payment → gets the result. Fully automated, no human needed.
For developers who prefer traditional billing, we also support fiat top-ups via Stripe.
Pricing
No minimum. No subscription. Pay only when you use.
| Tool | Price |
|---|---|
| Image generation (Flux 1.1 Pro) | $0.005 |
| Web scraping | $0.002 |
| Code execution (E2B) | $0.005 |
| Browser automation | $0.005 |
| Audio transcription | $0.003 |
| Text-to-speech | $0.001 |
| Web search | $0.001 |
| Weather, QR, crypto, DNS | $0.001 |
$1 = hundreds of API calls for most tools. A typical AI agent workflow (search + scrape + image + TTS) costs about $0.01.
MCP Support
IteraTools ships as a hosted MCP server. Add it to Claude Desktop, Cursor, or any MCP-compatible client:
{
"mcpServers": {
"iteratools": {
"url": "https://api.iteratools.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_KEY"
}
}
}
}
All 35+ tools are immediately available as MCP tools in your AI assistant.
What's Next
-
/sheets/readand/sheets/write— Google Sheets integration -
/email/send— Transactional email via Resend - More sandbox languages (R, Ruby, Go)
- Usage analytics dashboard
Try It + Support Us
🔗 iteratools.com — Get an API key, start calling tools
👍 Product Hunt — Vote for us today (March 10th!)
Questions about the architecture, security model, or pricing? Drop a comment below — I read everything.
Top comments (0)