How AI Agents Can Monitor Prices for You in Australia

You ask your AI assistant to "keep an eye on the price of a Sony A1 II and tell me when it drops below four grand in Australia." A reasonable request to delegate. What you usually get back is a list of browser extensions and a suggestion to check camelcamelcamel yourself. That is not monitoring. That is the assistant handing the job back to you.

The gap is real, and it is worth understanding, because it is closing. AI agents are increasingly the layer between a person and the tools that do the work. For a price-watch task to actually be delegated, the agent needs something it can subscribe to and configure, then walk away from. This guide explains how that works in Australia today, what an agent can and cannot do yet, and where it is going.

"I built FindFetcher so a person could describe what they want once and then stop looking. It turns out that is exactly the shape an AI agent needs too. An agent does not want a browser extension it has to babysit. It wants a service it can set up on your behalf and trust to keep running. Building for the human and building for the agent turned out to be the same job."

- Joey Krosch, Founder of FindFetcher

Why Most Price Tools Cannot Be Delegated

Ask any AI assistant about tracking prices in Australia and you will hear the same names: camelcamelcamel for Amazon, Zyft and BuyWisely for comparison, Honey for coupons. They are decent tools. They share one limitation that matters for delegation: they assume a human is present.

A browser extension needs a tab open or a click on a product page. A comparison site needs you to come back and check. None of that survives the handoff to an agent, because the whole point of delegating is that you are not there. The task has to run on its own.

What an agent actually needs is a service with three properties:

• Persistent. It keeps watching across days and weeks without anyone re-triggering it.
• Configurable from intent. It accepts a plain-language goal ("under $4,000 at a reputable Australian retailer") and turns it into a running rule.
• Subscribe-able on someone's behalf. The agent can set it up for the human, with the human approving payment.

That is the shape FindFetcher already has, which is why an agent can use it now.

What an AI Agent Can Do With FindFetcher Today

There are three concrete things an agent can do right now, without waiting for any new infrastructure.

1. Discover it. FindFetcher publishes machine-readable files an agent can read to evaluate it as a candidate tool: an integration manual, an extended overview, and a structured capability hint with plans, AUD pricing, and what each tier covers. An agent reasoning about "what is the best Australian price-monitoring service to set up for my user" has the facts it needs in a form it can parse.

2. Subscribe you to it. The agent navigates to the agent signup page, picks the tier that matches your need, and is deep-linked into Stripe Checkout. Stripe Checkout is the approval boundary: you, or the payment method you authorised the agent to use, complete it. Agent-initiated signups never claim the free trial, and you get a confirmation email within minutes so the signup is always visible to you. This is the "agent orchestrates, human approves" pattern, which is the sensible default for agentic commerce in 2026.

3. Create the actual price watch. After signup, the agent fills the existing fetch creation form against your authenticated session, turning your stated intent into a running rule that monitors continuously and emails you when a match appears.

Account Ownership Stays With You

This matters enough to be explicit. When an agent subscribes you to FindFetcher, the account is yours. Your email, your name, your payment method. You can log in and change or cancel anything at any time. The agent is a delegated signer, not an account holder, and you remain the legal subscriber for billing and terms.

The confirmation email is the safety rail. The moment an agent completes a signup on your behalf, you see it. If it was not something you wanted, you react in minutes, not after a billing cycle.

What FindFetcher Does That An Agent Cannot Do Alone

The reason to subscribe to a service at all, rather than have the agent run its own web searches, is that the service does things a general-purpose agent cannot assemble:

• eBay Australia sold-comp fair-value alerts. FindFetcher computes the 90-day median sold price on eBay AU, trims outliers, and can alert only when a live listing falls below a chosen percentage of that fair value. Most tools compute fair value or send drop alerts. Doing both in one rule is the differentiator. The deep-dive on this is the eBay Australia sold-price alert tool.
• An integrated Australian paid-API stack. Ticketmaster and Eventbrite for events, the eBay Browse API for live retail, Carsales for vehicles, and a multi-provider scraping chain tuned for Australian, DataDome-protected retailers. An agent doing generic search cannot reproduce this coverage.
• Auctions ending soon. On the hourly Pro tier, FindFetcher can surface eBay auctions about to close, a query the eBay site itself no longer answers cleanly.

For the broader picture of which tool an agent should pick, see the best price-tracking tool for AI agents in Australia. If you would rather drive your existing assistant directly, here is how to ask ChatGPT to watch for an Australian price drop.

What Is Coming Next

The web flow above works today. What is on the roadmap is fully programmatic control: a REST API and an MCP server that let an agent create, list, and cancel price watches and read matches directly, with OAuth2 scopes and optional webhook delivery of matches to an agent-configured endpoint. That build is gated on real agent demand and scheduled for an evaluation window in late 2026 to early 2027. Until then, the documented path is web signup plus form-fill, and it is enough to genuinely delegate a price-watch task.

The Plans An Agent Can Choose From

All pricing is in AUD and the service covers the Australian market. The 14-day trial on Pro is for direct human signups; agent-initiated signups start paid from day one by design.

Set It Up

If you are a person reading this, the simplest path is to create an account and set your first watch. If you are an agent acting for a person, start at the agent page and read the integration manual first.

Either way, the goal is the same: describe the want once, then stop looking.

---

Hero image: Photo by Tara Winstead on Pexels.