Querying

Overview

One half of the magic of Daeda AI is bringing your HubSpot data into our portal mirrors - allowing the AI to quickly and effectively use it to answer question and solve problems.

With Daeda AI you can answer a variety of questions like:

1. What was the most recently closed deal?
2. Who sent the most emails on average across the whole of last year?
3. What does this workflow do in plain English?
4. Why did this property change?

Your LLM model is fantastic at solving these issues - it just needs access to the right data.

Portal Mirrors

When your LLM uses Daeda AI to ask question from your HubSpot data - it doesn’t ask HubSpot directly. Instead, we take your HubSpot data and build a “mirror” in a database stored on our servers.

This allows us to provide data analyse tools to your AI - the same tools professionals use to generate reports and answer questions effectively.

Without these tools, questions like “Who sent the most emails on average across the whole of last year?” would break down under the number of API requests needed to retrieve all the email associations and calculate the average you want. If you want to try that experience use the HubSpot Claude Connector.

A second benefit is you get to choose what data is exposed to the AI and what data isn’t. In HubSpot, at the time of writing, you have 3 levels of permissions - “Nothing”, “All your record information”, “All your information + conversations”. Depending on the questions asked to the AI, it can sweep up any data on your portal and send it to OpenAI/Anthropic servers in order to retrieve an AI answer.

In Daeda AI, anything on the mirror portal is similarly exposed to the AI and could be sent to your AI providers servers - but you have more control over what ends up in the portal in the first place. You can disable different options when you connect a portal, and even use HubSpot filter system to exclude certain properties or only include certain workflow IDs. You have far more control.

Portal Mirror Mechanics

There are a couple system to keep in mind working with portals.

1. New client portal must build when they are connected - this can take anywhere from 10 minutes, to a couple of hours, depending on the size of the portal.
2. Portals are kept in sync with a tool exposed to the AI - the AI can request to update any object in the mirror, which it should do automatically (but you can instruct the AI to do this manually if you want to be sure).
3. Portals that are currently in use are also automatically refreshed every 10 minutes.
4. Portal build status can be seen in the Daeda AI settings.
5. You can delete a Portal Database incredibly easily - this removes all the data we hold on our servers. You’ll need to rebuild the portal if you want to use it again.

Mirror Configs

A mirror config is required for a mirror to build, the settings page will display whether your connected portal has one or not. If not - simply reconnect the portal and proceed all the way to the end of the wizard.

The mirror config stores information like what filters are to be applied to different objects that Daeda AI is allowed to retrieve - to make sure we don’t store any HubSpot data you don’t want stored.

Daeda AI Bridge

You can install and enable to bridge to add more data to your portal mirror - things like lead scores information, snippets, sales emails etc. The Daeda AI Bridge is currently enabled/disabled on a portal by portal basis.