Introducing Notehub IQ: Use AI to Talk to Your Products

Your Notecard-powered product, with fleets of devices in the field, has been reporting location and sensor data to Notehub for months. Then one day someone on your team asks a very reasonable question:

Did any shipment sit above 8°C for more than 15 continuous minutes last week, and if so, which devices were they?

Of course, the data to answer that is already in Notehub and your own repositories, but getting to it is the hard part. You're likely routing events to your own cloud app, but to truly query it you need to stand up a database, define a schema, write enough SQL to handle time windows and gaps, and maybe publish a dashboard you look at on occasion. Then you realize this single straightforward operational question has turned into a data-engineering project, and you repeat some version of that work every time the question changes.

This is exactly why we're pleased to announce the public beta of Notehub IQ, a new way to ask questions of your products in plain language, and get answers grounded in real device data.

tl;dr Notehub IQ is an intelligence layer for Notehub. It pairs a first-party storage layer with the Notehub MCP (a Model Context Protocol server that your existing AI agent connects to) so you can ask your products a question in plain language and get an answer from the data you already collect. There's no pipeline to build and no data warehouse to feed (and it's completely free while in public beta!).

What Exactly is Notehub IQ?

Today, Notehub IQ is comprised of two Blues-provided components, used with the AI agent you already have:

1. A Blues-provided storage service acts as the first-party, analytics-ready data layer. You choose which Notefiles to include, and the mirrored storage service automatically stays in sync as new events arrive.
2. The Notehub MCP is the interface. It's the part your AI agent talks to, and it's responsible for turning your question into a valid query.

You use Notehub IQ through your own AI agent, like Claude Desktop or Web (our recommended path), Claude Code, Codex, VS Code Copilot, Cursor, and others.

Because the storage is first-party, there's no separate warehouse to feed and no export job to babysit. Your data is already in Notehub, and Notehub IQ makes it queryable in place, whether you have a hundred devices or a million.

How It Works

With Notehub IQ enabled, when you ask a question of your products, your AI agent will hand it to the Notehub MCP. The MCP will inspect your datasets (your Notefiles that you have onboarded), write the SQL query, run it against the storage service, and pass the results back to the model to answer in plain language.

Two things are worth calling out here:

First, every query is read-only. Notehub IQ can read and analyze your data, but it can't change, move, or delete any of it.

Second, it understands how Notehub actually behaves. For example, a gap in the received_at timestamp in a series of events doesn't mean a device went offline, and Notehub MCP knows that. It also knows to look at _session.qo for connectivity instead of counting raw events, which is the difference between "this device is dead!" and "this device is fine and happens to sync every six hours!"

The Kinds of Questions You Can Ask

The options are (literally) endless, and every product is different, but here are some example questions you can ask of your product data:

1. "Which devices are potentially disabled and not responding (or are they just mid-cycle and syncing on their normal schedule)?"
2. "Did any temperature reading go under 0°C for more than 5 continuous minutes last week?"
3. "Which devices are using the most cellular data, and is payload size or sync frequency driving it?"
4. "Rank my solar/battery-powered units by how close they are to a brown-out."

As a personal example, I've been testing the Notehub MCP on a project I wrote up some time ago, The Three Fridge Problem. I had a small fleet of refrigerators to monitor, everything deployed locally. To simulate an anomaly, I manually set the location of one device to be far away from my "home base". So I asked...

Are all my fridges where they're supposed to be?

Interesting, right? However, the real value of the Notehub MCP shows up when it's one of several tools your AI agent can reach. If Notehub is the source of truth for your products, when you point your agent at the other connectors it already has (e.g. Slack, a ticketing system, or any number of third-party APIs) it can go from answering a question to doing something useful with the answer.

For example, Claude Desktop's scheduled tasks can run a prompt on a schedule, unattended. Imagine a routine that runs every morning at 7am: it asks the Notehub MCP whether any device has drifted out of place, crept above its temperature threshold, or gone quiet in the last 24 hours. If something is off, it posts a short summary to your team's Slack channel and opens a ticket in your support queue with the offending device IDs pre-filled.

You've turned a manual action you had to remember to perform into a check that runs itself and finds you where you already work.

Getting Started

Notehub IQ is in public beta today, and it's free to use during the beta. To try it, point your MCP client at the Notehub MCP:

https://mcp.notehub.io/mcp

Your client will prompt you to sign in to Notehub and authorize access the first time it connects. From there, a typical first session looks like this:

1. Tell your AI agent which Notehub project you want to use.
2. Tell your AI agent which of your existing Notefiles (e.g. sensors.qo, data.qo, etc) you want to mirror in new datasets.
3. Your AI agent will then let you know when your datasets are ready. From there you can start asking questions!

The Notehub IQ documentation covers client setup and the full list of available tools. If you're already building on Notehub, this is a no-brainer addition to your workflow. And because you just ask in plain language, anyone on the team can use it, no data specialist required.

We can't wait to see what you learn about your Notecard-powered product with Notehub IQ. 💙

Frequently Asked Questions

What is Notehub IQ?

Notehub IQ is an intelligence layer for Notehub. You ask questions in plain language from your own LLM Client or AI agent, and the Notehub MCP gives that agent access to your real device data so it can answer them.

Do I need to know SQL to use Notehub IQ?

No. You ask questions in plain language and the Notehub MCP writes and runs the SQL for you. Knowing SQL helps if you want to review or refine what it generates, but it isn't required.

Can Notehub IQ change or delete my data?

No. Every query the Notehub MCP runs is read-only (SELECT only). It can read and analyze your data, but it cannot modify, move, or delete any of it.

Which LLM Clients and AI agents does Notehub IQ work with?

Any MCP-capable client, including Claude Desktop, Claude Code, Codex, VS Code Copilot, and Cursor. You point it at the Notehub MCP at https://mcp.notehub.io/mcp and sign in with your Notehub account.

How much does Notehub IQ cost?

Notehub IQ is free to use while it is in public beta. Pricing beyond the public beta is coming soon.