Get started with Recast MCP now
We just shipped two new ways to use Recast inside the AI assistants your team already works with. Add our plugin to Claude in a single click, or connect over MCP from ChatGPT, Codex, or anywhere else. Either way, it’s the Recast engine itself in the hands of the LLMs you use every day.
It means you can ask questions in plain English, get answers straight from your live models, do powerful analysis, weave in external data, automate the pulls, and so much more. Here’s a taste of what you can do:
- Rank every promo you ran last year by the real incremental value they added by running the Spike Detail report.
- Join Recast’s incremental ROl estimates with your own Google Ads or Meta export and note where they diverge.
- Loop Spend Response Curves across all of your channels and rank them by how much headroom each has left before it saturates.
- Trace every lower-funnel channel back to the upstream spend that fed it with the Lower Funnel Waterfall, then rank your upper-funnel channels by the downstream revenue they created.
- Rank channels by marginal ROI and push the results to Slack for your team to discuss.
… and much more.
Today, Recast MCP provides access to 20 of our proprietary reports. Access to more tools is coming very soon.
What exactly is Recast MCP?
MCP stands for Model Context Protocol, the standard that lets AI assistants connect to outside tools (think Notion, Slack, Google Drive, etc.) and data sources. It’s the same idea as any integration between two platforms, except the thing on the other end is your Recast model. Our Claude plugin is simply the easiest way to switch that connection on inside Claude. It bundles the same MCP setup into a one-click install, so you can skip the manual configuration.
Here are a few more things Recast MCP enables:
- Ask in plain English. A non-technical marketer can now ask “once we set aside the purchases that would have happened anyway, which of this year’s promotions added the most real revenue?” right inside their AI assistant, with no dashboard and no SQL.
- Blend Recast with your own data. An analyst can join Recast’s incrementality estimates with data from Google Ads, Meta, or other first-party data in a single workflow, so the model’s results sit right next to everything else their business reports on.
- Automate the weekly pulls. Recast refreshes your entire model every week, thousands of parameters re-estimated. The Claude plugin lets teams schedule those pulls straight into their own agents, Python/R scripts, and dashboards, instead of re-exporting them by hand every Monday.
Getting started
The Recast MCP and Claude Plugin are now available to all Recast customers. Connecting runs over a standard OAuth browser sign-in, and once you’re provisioned, it takes about two minutes to wire up Recast from Claude, Codex, or the AI client of your choice. If you need any assistance, just reach out to your marketing data scientist.
An important note: your data stays yours. Recast never uses it to train generative AI models, and it flows over an encrypted connection only to the AI client you authorized through OAuth.
This is another step toward Recast living wherever your team already works. We are expanding what you can do through the MCP over the coming months, so more of the platform will quickly become reachable from inside your AI tools.
We’d love to hear how you put it to work!
