# Agentic Automation to remove Cloud friction > Source: > Published: 2026-10-07 00:00:00+00:00 [Blog](https://medusajs.com/blog) October 7, 2026·Company # Agentic Automation to remove Cloud friction Shahed Nasser Shahed Nasser Learn how we use AI agents to connect user data, understand churn, and turn those insights into a better Medusa Cloud experience. At Medusa, we're automating our internal operations with AI to increase our efficiency. We're applying this across engineering, documentation, sales, and other operations. This has allowed us to dedicate our resources to shipping features faster for our users. So, we're starting a series of posts around the different agentic automations we have built. These posts will share how these processes work and our learnings from them. Check out the previous posts in the series: [up-to-date skills and MCP server](https://medusajs.com/blog/agentic-automation-skills-and-mcp-that-evolve-with-our-product) and [automated releases and changelog](https://medusajs.com/blog/agentic-automation-automated-releases-and-changelog). ## Churn as a source of feedback While we have almost zero churn from customers live on Medusa Cloud, some users churn in their early days as they get started. Here, churn is a valuable feedback mechanism for improving Medusa Cloud and understanding friction points. When users cancel their subscription, we ask them to share why. While not everyone provides detailed feedback, the insights we receive help us understand where users encounter friction and how we can better meet their needs. This was especially useful in the early days of building Medusa Cloud, when we were learning how users experienced the product. Understanding why users canceled helped us identify gaps and prioritize improvements, creating a better experience for both future and returning users. ### Analyzing the Data Just like any product, we have data spread out across different services and databases related to how our users are using Medusa Cloud services. This includes: 1. **Metabase:** General user data, build and deployment statuses, and agent sessions users have with our[Cloud Assistant](https://docs.medusajs.com/cloud/assistant) . 2. **PostHog** : How our users browse our products, what errors they run into, and their[MCP server usage](https://docs.medusajs.com/cloud/medusa-mcp) . 3. **Linear:** Support tickets that users have opened and how we've resolved them. If you put all this data together, you can understand the user's journey better and identify issues they may have run into. However, doing this manually is complex and takes up a lot of time, especially if you want to find common patterns of why users are canceling. You need to: 1. Map these data into a uniform output. 2. Build linear user journeys that piece together all this information. 3. Contextualize the data with Medusa Cloud-specific updates and information. 4. Cluster the results across users and formulate hypotheses. ## How we Solved This with Agentic Automation Early in the year, I set out to solve this problem with Claude Code. I focused on putting together a user's journey across the different data sources, understanding all their friction points, and formulating a hypothesis of how we can improve Medusa Cloud for that user. Then, I applied that same logic across users within a time range and was able not only to better understand users individually but also to see patterns across plans, types of ecommerce stores, or integrations they're building. ### Subscription Cancellation Analysis Skill Once I reached a point where I trusted the flow to provide me with a useful analysis, whether for a user or a cluster of users that canceled within a time range, I asked Claude Code to create a skill for this analysis. This skill includes: 1. Information on where all the data lives, and how to connect to it. For Metabase, we connect with its APIs. For PostHog and Linear, we use their MCP servers. 2. What data are needed per organization for performing the analysis. This includes details like the organization's subscription plan, their MCP usage, what their failed build rate was, whether they've had any unresolved support requests, and more. 3. How to combine and cluster the results across organizations. It specifies the stats to calculate with all organization data and how to formulate a hypothesis based on this data. For example, focusing on whether the subscription cancellations were mainly of a specific plan, or whether the organizations shared the same friction points. ### Result: Full Report with Agentic Automation Now, we can use the skill to ask Claude Code to analyze Medusa Cloud subscription cancellations for a period of time, such as the past two weeks. It writes a general report with the summary and hypothesis it has, along with reports for specific organizations that are interesting to look into specifically. We don't rely on the report as-is. We often ask Claude Code for more details about certain aspects that raise our curiosity, and Claude Code amends results based on what it finds. The findings eventually lead us to action points that improve our products. For example, through this analysis we found that new users struggle to get a new project running on Medusa Cloud. So, we added an [onboarding flow](https://docs.medusajs.com/cloud/first-project) to make the process easier. ### Integration with Slack We've taken this a step further by integrating our custom agent, built on top of Claude Code, into our Slack channel. When an organization cancels its subscription, the agent automatically uses the skill to share a report of the organization, including what went wrong and potential friction points. This is useful to detect organizations that have canceled due to an immediate shortcoming on our end, such as a bug in our product. Then, we reach out to the organization owner to resolve the issue, which often leads them to re-subscribe. ## **Get Started with Medusa** If you're new to Medusa, check out the [documentation](https://docs.medusajs.com/learn), where you'll get a more in-depth understanding of what Medusa is, the commerce features it provides, and how to deploy Medusa to [Cloud](https://medusajs.com/pricing/).