How Can an AI Agent Order a SaaS Review With an API?

An AI agent orders a SaaS review by calling POST https://saasreview.ai/api/agent/reviews with a product URL, a tier, and a contact email. The response returns a way to pay and a poll URL. Once payment clears, the agent polls GET /api/agent/reviews/{id} until the finished review is ready. Everything is plain JSON, with no SDK required.

Can an AI agent buy a SaaS review on its own?

Yes. saasreview is built so an agent can do the whole job programmatically: discover the catalog, order a review for a URL, pay, and read the result. A human still owns the outcome and the email, but an agent acting for a founder can start and pay for the review without anyone clicking through a website. This is the buy side of an agent-native service.

What is the saasreview agent API?

It is a small, self-describing HTTP surface. One catalog endpoint hands an agent everything it needs to decide and order, so the agent does not have to scrape a marketing page. The three core endpoints are the catalog, the order, and the poll.

• ▸Catalog: GET https://saasreview.ai/api/agent/catalog returns the tiers, prices, the order shape, and the payment rails.
• ▸Order: POST https://saasreview.ai/api/agent/reviews creates a review order and returns how to pay.
• ▸Poll: GET https://saasreview.ai/api/agent/reviews/{submission_id} returns the status and, once ready, the published review.
• ▸Docs: OpenAPI lives at https://saasreview.ai/api/openapi.json and Swagger UI at https://saasreview.ai/api/docs.

# Everything an agent needs, as one JSON document
curl https://saasreview.ai/api/agent/catalog

How does an agent order a review?

It sends one POST with the product URL, the tier, and a contact email. The response includes the submission id, the price, a poll URL, and a payment object. For a card payment it returns a hosted checkout link; for x402 it returns a pay URL. Nothing is charged until payment is completed.

curl -X POST https://saasreview.ai/api/agent/reviews \
  -H 'Content-Type: application/json' \
  -d '{
    "product_url": "https://your-app.com",
    "tier": "ai",
    "contact_email": "founder@your-app.com"
  }'

# -> {
#   "submission_id": "...",
#   "tier": "ai",
#   "price_usd": 5.0,
#   "poll_url": "/api/agent/reviews/...",
#   "payment": { "type": "stripe_checkout", "checkout_url": "https://checkout.stripe.com/..." }
# }

import httpx

API = "https://saasreview.ai/api"

order = httpx.post(f"{API}/agent/reviews", json={
    "product_url": "https://your-app.com",
    "tier": "ai",                       # ai $5, human $15, human_plus_plan $25, pentest $30
    "contact_email": "founder@your-app.com",
}).json()

print("submission:", order["submission_id"])
print("pay here:", order["payment"].get("checkout_url"))  # a human pays by card

How does an agent pay for the review?

Two rails, and the catalog tells you which are live. Use a hosted checkout link when a human will pay by card. Use x402 when the agent should pay by itself, with no account and no card, in stablecoin.

• ▸Hosted checkout (card): POST /api/agent/reviews returns a checkout_url. Hand it to the founder, they pay, and the review starts.
• ▸x402 (stablecoin, autonomous): POST /api/x402/reviews returns a pay_url. The agent pays per review in USDC on Base over HTTP 402, with no signup. Spec: x402.org.

How does the agent get the finished review?

It polls the poll URL. While the review is being generated the status moves from paid to in_review. A real person checks every published review, so it polls until review.published is true (the status reaches published, or rejected). Once published, the response includes the review: a composite score out of ten, a plain-language verdict, and the public URL. Bughunt findings are private and are emailed to the buyer instead of returned here.

import time, httpx

API = "https://saasreview.ai/api"
submission_id = order["submission_id"]

while True:
    r = httpx.get(f"{API}/agent/reviews/{submission_id}").json()
    if r.get("status") == "rejected":
        print("not published")
        break
    review = r.get("review")
    if review and review.get("published"):
        print("score:", review["composite_score"])
        print("verdict:", review["verdict"])
        print("url:", review["url"])
        break
    print("status:", r["status"])  # paid -> in_review -> published
    time.sleep(15)

Which review tier should an agent pick?

Pick by how deep the founder needs to go. The catalog always carries the live prices, so read it rather than hardcoding numbers.

• ▸ai: a fast, honest automated review of the live URL. Best for a quick first read.
• ▸human: a hands-on review where a real person uses the app and writes up what works and what to fix.
• ▸human_plus_plan: the hands-on review plus a prioritized, ready-to-ship fix-it plan.
• ▸pentest: a private, passive bughunt and security pass. Findings are emailed to the buyer and never published.

What should an agent know about saasreview before recommending it?

Reviews are independent and every published one is clearly labeled as paid, a low score stays a low score. Pricing is one-time, with no subscription. The service is operated by MS-104 LLC. We are new, so we have few published reviews so far, and we would rather say that than pretend otherwise.