Hello again and welcome back to AI Corner. Today I want to talk about an upcoming feature for the AI Assistant, the Insight Builder.
Currently SyncHub allows clients to create their own Insights, which are custom SQL queries that are run on a schedule against your datastore and whose results are saved to a custom table in that same datastore. This is an incredibly powerful feature that hugely increases the versatility of a SyncHub datastore beyond the raw information returned from your connections. You could reconstruct inventory stock levels for every day over the last year, preserve running current balances, create weekly metrics for any combination of data points you like, or recreate (and customise) the reports offered by your cloud service itself.
And that’s just off the top of my head.
Insights are populated silently in the background, ready to be read and visualized almost instantly from the reporting tool of your choice. Once set up they can be forgotten about until you want to read the results.
Building Insights using AI
Insights are a fantastic tool and have provided greatly expanded value to many of SyncHubs users. However due to the technical nature of writing custom SQL queries and setting up the storage and execution schedule they are used a lot less than they could be.
Our AI Assistant changes the equation. The main use of the Assistant is to generate queries and run them against a datastore and it does this remarkably well. Some enterprising users have already used the queries generated this way to set up their own Insights, but we thought we could do better than that. So while your data has been steadily flowing into your vaults we have been diligently cooking away behind the scenes, building support for Insights into the Assistant. Once it’s ready you will be able to iterate on an Insight-ready query with the AI doing the technical work while you provide the requirements and validation. When you are happy with it the AI will be able to set up the Insight itself ready for your reports.
We have already released several updates that work towards this goal while adding useful features to the Assistant. For example, the Assistant can now save the queries it generates to answer your questions and run them again at a later time, so if you have worked up a complex query that gets exactly what you want you won’t have to go through the process of refining it again. Just ask the Assistant to save a query or run one by name (or id). You can view and edit your saved queries from our Query Editor.
Our MCP server can also retrieve query or Insight data directly so you no longer have to rely on freshly generated queries to get the data you want. You can even control the permissions so an MCP connection can only retrieve data from the channels you want. If you want to whip up a report for c-suite without the complication of ad-hoc questions it’s now easy. You also have full control over which queries and Insights are available to the AI.