Ordergroove Agents are AI-powered specialists that run the ongoing work of your subscription program. They recover failed payments, catch churn early, and watch your performance data, continuously testing what works so you can act on it and grow faster—running more tests, and learning from each one, than a team could by hand. Payment, Retention, and Analytics are the first three, with more on the way.
This article is an overview, it explains what an agent is, how an agent differs from "just AI," and what our MCP connector does. From here, you can jump into the dedicated article for each agent to see how it works and how to turn it on.
Feature Access: Ordergroove Agents are currently in Early Access, with a limited number of spots. Fill out the Agents Form if interested. Requesting access doesn't guarantee a spot during Early Access, but it puts you on our radar as we expand.
What are agents?
An agent is an AI system that acts on your behalf: it answers questions, surfaces information, and then takes the next step. The simplest way to think about it is a chatbot tells you what to do, while an agent does the work.
That's also the difference between an agent and "just" AI:
- Just AI can analyze data, generate content, or make a prediction. It might tell you, "this subscriber looks likely to cancel."
- An agent takes the next step. It builds the retention workflow, sets up the payment retry, or drafts the campaign - and then either acts on it or hands it to you to approve.
A helpful way to picture it: if AI is the brain, an agent is the brain plus the hands.
Note: Agents are powered by large language models (LLMs), the same kind of model behind tools like Claude and ChatGPT. The LLM is the reasoning engine; the agent is what turns that reasoning into action inside Ordergroove.
What is MCP?
MCP stands for Model Context Protocol. Think of it like a bridge for connecting your internal data or tools to an AI client like Claude or ChatGPT. MCP lets your own AI ‘talk’ to your Ordergroove data and answer questions using it.
- An agent is the thing doing the reasoning and acting.
- MCP is the secure connection that lets an agent reach your Ordergroove data. It’s like a bridge or pipeline, moving data from one place to another.
So when you connect your AI client to Ordergroove's MCP, your assistant can work with your Ordergroove data. Today, that access is read-only - your AI client can ask questions and pull data, but can't change anything in your program. You can learn more in the Ordergroove MCP article.
The Ordergroove Agents
There are two ways to put agents to work with your Ordergroove data:
- Ordergroove's own agents, built right into the Ordergroove dashboard. These run against the data Ordergroove already has, with little to no setup on your end.
- Your own AI client (Claude, ChatGPT, or another MCP-compatible tool), connected through Ordergroove's MCP. This brings your subscription data into the tools your team already uses.
Ordergroove offers three agents today, with more coming in the future. Each one shows you what it's doing, and additional guardrails and merchant controls are coming in future stages:
Payment Agent
The Payment Agent works behind the scenes to recover more revenue from failed payments. It runs controlled A/B tests on different retry strategies against your current setup, automatically scales the approaches that win, and then starts looking for the next improvement. During Early Access it focuses on payments that fail due to insufficient funds. It runs on its own, and gives you a clear log of every test it has run and what happened. See the Payment Agent article for details.
Retention Agent
The Retention Agent looks for patterns in where subscribers drop off—across different product and frequency combinations—and surfaces the opportunities worth acting on. When it spots one, it hands you a pre-built workflow aimed at exactly that group of subscribers.
Every workflow it recommends is set up as an A/B test against your baseline, so you can measure the strategy's real impact rather than guessing. You review it, add the incentive or gift you want to offer, and approve it before anything goes live—nothing the Retention Agent suggests is activated without your sign-off. See the Retention Agent article for details.
Analytics Agent
The Analytics Agent helps you understand what's happening in your program without the manual digging. You can ask questions of your data in plain language — "which products drive the most subscription revenue?" — and get back answers, tables, and visualizations. At launch, the Analytics Agent's capabilities are available through Ordergroove's MCP connector, which you connect to your own AI client; a version built natively into the Ordergroove dashboard is coming soon. See the Analytics Agent and Ordergroove MCP articles for details.
FAQ
How do I get access to Ordergroove Agents?
Agents are in Early Access. Please reach out to your Customer Success Manager or Ordergroove Support, and we'll help get you set up.
Which agents are available right now?
The Payment Agent and Retention Agent are available in the Ordergroove dashboard during Early Access. The Analytics Agent's capabilities are partially available now through our MCP connector, with a native in-dashboard experience following shortly after.
Do I need to do any setup or technical work?
For the agents that run inside the Ordergroove dashboard, there's almost no setup—they work against data Ordergroove already has. Using your own AI client through MCP takes a one-time connection in that client (for example, adding Ordergroove's MCP to Claude), and our team will help you through it.
Can I use my own AI tools?
Yes. Through Ordergroove's MCP connector, you can connect Claude, ChatGPT, or another MCP-compatible client and work with your Ordergroove data right alongside the other tools your team already uses. See the Ordergroove MCP article.
Does this work on my eCommerce platform?
Ordergroove's architecture is platform-agnostic, so agents are designed to work regardless of your storefront. A few capabilities have platform-specific limits during Early Access—for example, the Payment Agent currently requires Shopify. The dedicated article for each agent has the current details.
What if I disagree with something an agent suggests? Can I override it?
Yes. The Retention Agent and Analytics Agent are advisory—you can dismiss any insight or decline any workflow you're not interested in, and nothing happens unless you approve it. The Payment Agent runs its tests on its own during Early Access, but it shows you exactly what it tested and what the results were.
Can I see what the agents have done?
Yes. Each agent gives you visibility into its work: the Retention and Analytics Agents keep a feed of what they've surfaced, and the Payment Agent logs every test it runs—including the results and a plain-language explanation of which strategy it chose and why.
Which AI model do the agents use?
Ordergroove's agents are model-agnostic—we continuously test across leading models and pick the best fit for each task, so you don't have to manage any of it. If you'd rather use a specific model yourself, you can do that through MCP with your own AI client.
Additional reading
- Payment Agent — how the Payment Agent recovers revenue from failed payments.
- Retention Agent — how the Retention Agent spots subscriber drop-off and builds workflows for your approval.
- Analytics Agent — ask questions and get insights from your program data; today it runs through this connector.
- Ordergroove MCP — how to connect your AI client to your Ordergroove data, and the full list of data you can ask about.