Discover how Click Rain uses Model Context Protocol (MCP) servers to bridge AI platforms like Claude with core business tools—such as Harvest, Wrike, Bitbucket, and Craft CMS. By unlocking siloed data, MCP enables teams to streamline project status tracking, automate website QA, enhance content management, and improve project estimation accuracy.
The swift advancement of AI models and platforms has promised transformative change for organizations, both at the enterprise level and for small businesses. During the past year, we’ve focused on practical opportunities at Click Rain to utilize this rapidly changing technology, with some of the most immediate gains originating from Model Context Protocol (MCP) server connections.
Model Content Protocol (or MCP) servers are the communication bridge between AI platforms such as Claude or ChatGPT and the systems or software that you may be using to run your business. At Click Rain, we use a wide variety of platforms to assist in operating our marketing agency. A few examples of these include the following:
Bitbucket, for our development team to manage code repositories
Figma, for our creative team to craft delightful website designs
Harvest, for our operations team to track hours and billing
Wrike, for our account team to flight out tasks and projects
Google Workspace, for our organization to manage documents, spreadsheets, and presentations
Slack, for our team members to communicate asynchronously
Each platform serves a defined purpose but data can often end up siloed. MCP servers help unlock this data to help answer organizational questions and link disparate systems together.
This has been a year of dreaming at Click Rain, with AI opening up new and exciting ways to solve problems and improve processes. We’ve found ourselves regularly asking the question, “What if…?”
What if we could provide a high-level website project status that summarizes data from several different systems?
What if we could improve our website QA process by identifying more issues and automating their resolution?
What if we could give tools for our team–and clients–to interact with our content management system in new ways?
What if we could reimagine our project estimation process?
In each of these “What if…” scenarios, we used MCP servers to help connect AI systems to our data and explore new ways to interact with it.
Each year, we build a lot of websites, with team members from our account, creative, development, content, and marketing teams convening to create innovative web solutions for our clients. These projects often span many weeks (or months) and touch a wide variety of systems throughout the project lifecycle.
This often makes determining a project’s full status challenging, with meeting notes living in Google Drive, project tasks managed in Wrike, budget tracked in Harvest, and impromptu team communication in project Slack channels.
Using MCP, we connected Claude to Harvest, Google Drive, Wrike, and Slack and created a Claude Skill to retrieve and summarize the project’s full status.
Claude returns a report that provides details on recent meetings, TODOs from Slack conversations, current budget status and/or overages, and project task next steps. This greatly reduces the siloes of each system, allowing team members to get the full picture at any time for any project.
Click Rain has long distinguished itself as the go-to agency for high performance–and high quality–websites. Our teams follow defined processes for identifying bugs, visual issues, accessibility concerns, typos, and other problems before each site is launched. We asked ourselves, “Is there a way to harness the power of AI tools to catch more issues–and resolve them?”
We created MCP server connections for Claude to connect to Craft CMS, Marker.io, and Bitbucket and set it to work. Claude crawls the staging site, using a Claude Skill to identify accessibility concerns, functionality problems, and content issues. It then reports these via Marker.io. Claude categorizes each item as either a code, design, or content issue and logs a full summary and screenshot for each entry.
For content issues, we can ask Claude to connect to Craft CMS and automatically resolve typos, broken links, images without alt text, and related problems. After resolving the related issues, each Marker.io entry is marked as complete with a summary of changes. Craft CMS logs the history of each entry, so we can easily revert to previous versions as needed.
This process has greatly improved our QA screening for each site and helps uncover issues that may have otherwise been missed.
Click Rain’s Content Management System (CMS) of choice is Craft CMS. We love Craft for its ease of use and extreme flexibility. Our clients love it, too.
We wondered, “What if there is a way to open up Craft CMS to AI tools in order to manage content in new ways?” Our team answered this question by building a robust MCP server plugin specifically for Craft, opening up a myriad of new opportunities.
After connecting Claude to the Craft CMS MCP server, we used the power of AI to automate traditionally challenging content problems:
Generating alt text for images for accessibility compliance
Migrating open ended content from external systems
Stubbing out initial site structure from a spreadsheet content plan
By utilizing this MCP connection, we can use natural language requests in Claude to ask questions about existing content–or build new content in the CMS.
A core business process at Click Rain is the website estimation process. Each week, team members gather to review new business opportunities and provide clients with estimates for their projects.
Click Rain built a custom web application called Tally to store and manage each client estimate. This system provides an interface to define project requirements and related costs. Over time, our Tally database has grown to represent hundreds of project estimates covering a wide range of budgets and scopes.
We asked ourselves, “What if we could connect Claude to Tally in order to better understand our estimates and more easily create new ones?” Using a custom MCP server connection, we can prompt Claude to review a client’s project request, compare scoped components to past projects in Tally, and recommend the related estimates. Our operations team can utilize this connection in conjunction with an MCP connection to Harvest in order to compare initial estimates to actual logged hours, helping improve estimate accuracy for future projects.
MCP server connections are helping our agency unlock data and improve processes in new and exciting ways–and we’re only scratching the surface.
Interested in using MCP servers in your organization, or have a system that could benefit from a custom built MCP connection? Get in touch! We’d love to talk about innovative ways to help improve your processes and systems.
Click Rain has been building MCP connectors to fill the gap for services that don't yet provide this important AI connection. If you have a custom MCP need, let's talk!
Brady is the vice president of operations at Click Rain Inc, leading company-wide operations initiatives. He is also Click Rain's foremost IT expert.
Read more »
Monthly digital marketing insights sent straight to your inbox, on us.