Public data inside your AI assistant: new MCP tools
For a long time, using public data meant knowing where the tables lived and how to pull them. That has changed. A new set of tools now lets you ask public data questions from inside the AI assistant you already use, and get a figure back in seconds. If you have ever wanted the median income of a county or the broadband access rate of a region without leaving a chat window, this is a real gain.
It also raises a fair question for anyone who builds audiences or personas. If an assistant can now reach census-grade data directly, what is a verified persona still for? Here are the main tools, what each is genuinely good at, and where a verified persona is a different job.
What an MCP actually is
The tools behind this all use the same standard. The Model Context Protocol, or MCP, lets an AI assistant such as Claude connect to an outside data source and query it live, rather than relying on whatever the model happened to memorise. It is the same idea behind Cambium AI's own Product Manager MCP, which runs research questions from a single prompt inside Claude. The difference between the tools below is what data they reach and what they do with it.
The public data tools worth knowing
Google Data Commons MCP.
Hosted on Google Cloud since February 2026, it lets an assistant query Data Commons, which brings together Census, World Bank, and other public sources, with no local setup. Best for a fast, broad factual lookup grounded in a trusted aggregate.
us-gov-open-data-mcp.
An open-source MCP server covering more than 40 US government data APIs across 250-plus tools, from the Treasury and the Federal Reserve's economic data to health and legislative sources. Best for developers who want direct breadth across government APIs. It is maintained by one person, so it is worth reading the commitment before you build on it.
OpenData (tryopendata.ai).
A free query layer across many government portals that ships as a Claude plugin and renders charts and tables inside Claude and ChatGPT. Best for a no-signup, broad first look at a public dataset.
General assistants with web search (Claude, ChatGPT).
Not public data tools as such, but with search turned on, they can fetch and summarise a published figure on demand. Best for a quick answer, as long as you check the source the assistant used.
What these tools are good at
All of them lower the cost of the do-it-yourself path. A marketer with a Claude account can now pull a population figure without opening a data portal or writing a query. For a headline number about a place, one of these is often the quickest route to it, and that makes public data more usable for far more people. They are a good addition to a marketer's stack, and they sit comfortably alongside persona work rather than replacing it.
Where a verified persona is a different job
Here is the honest distinction. These tools give you access to data. A verified persona is a different kind of work: it joins records across surveys at the level of the individual to produce a whole population at a real sample size, not a single number looked up in isolation. An assistant can tell you the median income of a county. What it cannot do on its own is hand you a representative population in which income, age, household size, language, and education move together the way they do in real people. That record-level joining is the gap, not always data access.
When to use which
Match the tool to the work. Need one figure on a broad topic, quickly? Google Data Commons MCP or an assistant with search. Building against many government APIs? us-gov-open-data-mcp or OpenData. Making a decision that carries value, a budget, a launch, a channel, or an audience call? That is when you want a population you can question and check, not a number in isolation.
Cambium AI meets these tools on the same ground and takes the next step. It turns the same verified public data into a synthetic population you can interrogate in plain English, down to the local level, from inside Claude or the product itself. An MCP that looks up a figure is genuinely useful. A persona you can trace back to the source, when money is on the line, is the insurance policy for a decision you cannot easily take back.
.png)
.png)