.png)
Last week we launched Count AI - a powerful extension of the canvas that allows you to generate, edit, and explain entire analyses, reports, and data models in seconds.
We know we’re releasing our AI features later than most, but it’s for a good reason. Not only did we want to make sure our AI features were secure, and well-designed, but with the unique UI of the canvas we had an opportunity to go beyond the standard AI chatbots and co-pilots.
The flexibility of the canvas combined with the power of multimodal models has allowed us to create what we think is a unique and game-changing analytical experience.
To best explain how Count AI works, we’ll dive into a few examples:
One of the most powerful uses of AI across many applications is its ability to speed up mundane tasks. For analytics, that means writing boilerplate or simple code for you.
In this example, Oli is using Count AI to generate a new SQL cell which does some basic data cleaning. With a short command, a new SQL cell is created:
You can use commands like these to generate any SQL cell, Python cell, visual, control cell, or as we’ll see later, any combination of those.
Count AI is also available on certain objects, like on a particular set of cells or objects. Continuing from the above data cleaning example, we can use Count AI to tweak the SQL that’s been generated:
This contextualization can become very powerful when applied to many objects at once, as we’ll see in a later example.
Editing visuals is one area we were particularly keen to tackle with AI. There are so many ways in which you might want to edit a visual - update a color scheme, change the axis range, change the axis labels, or scaling, change from a bar to a line chart, swap axes, etc - that it is often easier and much faster to type your commands rather than perform a series of menu clicks:
You can also apply the concepts of generating and editing objects at scale. Say, for example, you wanted to generate a multi-cell analysis. You might outline the analysis you want in sticky notes, then use AI to turn those sticky notes into live, interconnected cells:
This is where the power of AI in the canvas really comes into its own. Something like this can take 10-15 minutes, even for the best analyst, and with AI the code can be written in seconds.
Because there is no black box, the code generated is visible and editable in case you want to make changes or just want to see for yourself how it’s done something.
“I’ve been using Count AI for a while and it helps so much for the stuff that isn’t complex. It saves you typing out boilerplate SQL and lets you focus on the difficult bit - actually deciding what you want to see, how to break down complex problems, and how best to visualize it. That’s what makes a great analyst.” - Oliver Pike
Count AI can use images, sticky notes, drawings, text boxes, and even connections between objects as context to generate cells, visuals, and control cells. This means you can draw a graph or dashboard in the canvas using only the pen and text boxes, and generate the visuals behind it in seconds:
As a collaborative tool, it’s important that every Count canvas is easily understood by a wide range of users. To help with this, we’ve also added an Explain option to Count AI, which lets you select any part of a canvas, and have it explained in simple terms:
The explain function works on any combination of SQL cells, Python cells, visuals, and control cells.
We are launching Count AI initially for all paid accounts on July 15th. For our free accounts, Count AI will be available later this summer as we finalize cost assessments associated with the increased computational demands.
During the beta phase, you will have the opportunity to explore all these powerful features at no additional cost. We encourage you to provide feedback, share your experiences, and help us refine the AI capabilities further. Together, we can make data analysis more intuitive and efficient than ever before.
Thank you for being part of our journey. We’re excited to see how Count AI will enhance your analytics and can't wait to hear your thoughts. Stay tuned for more updates, and join our user community for ongoing discussions and support.
Q: Is Count AI secure?
A: Count AI is built upon Count’s secure data platform, and is fully GDPR compliant.
Q: Does Count or its partners train models using customer data?
A: No. You don’t need to worry that content from your canvases might appear in suggestions to other users.
Q: Is Count AI likely to lead to more queries running on my database?
A: No queries are run on your database without your approval. Count AI will be an opt-in feature for existing workspaces and can be disabled at any time.
Q: I have ideas for other uses of AI in the Count canvas, how do I share them?
A: We’re working on a way for Count AI Beta users to provide feedback, but if you have any before that, feel free to drop us a line at [email protected].
Q: How much will it cost?
A: There will be no change in pricing for those on annual plans. We’ll know more about the cost for other accounts after the Count AI Beta program.
