We know that formulas are a large part of Coda’s learning curve — that’s why we’re excited to launch two new content builders to help you get the most out of tables without needing formulas:
The compose column
Compose is a new column type that eliminates the need to rely on the Concatenate() and/or Format() formulas to create dynamic text in a table. Compose is a freeform editor that lets you write and format text, include references to columns using the @ symbol, and even formula results using the = symbol. The resulting dynamic content lives in a canvas and can be displayed like a normal canvas column.
You can even reference your compose column from other columns, which makes sending dynamic messages though a Gmail Send Message button or a Slack Post Message button fast and painless.
Earlier this year, we launched add related columns and the ability to show summaries on the canvas.
Now, we’ve made it even easier to summarize a table with its own slash command, /summarize table.
Table summaries surface key information using custom filters right on the canvas. You can get a count or list of rows from a table and summarize columns of all types. This lets you leverage formatting like callouts, headlines, and colored text to draw attention to the information your team needs to know – all without writing a single formula.
We can’t wait to see how you’ll use these new tools to power your docs. Have any other formula scenarios you’d like to see streamlined? Drop us a reply — our team’s listening in.
Seriously important feature, and especially for AI. AI needs content that is framed for the specific queries users will make. Table summaries especially, make this possible.
One of the challenges with using data in Coda AI is the limitation of the inference window (~8k and likely expanding soon). Despite this, large data sets cannot be included in prompts which means, we have to aggregate. ElasticSearch knows this drill well and one reason their data sets are effortlessly employed in AI.
Inclusion of subset data sets into AI blocks and other use cases, is now effortless thanks to this feature which was designed for human viewing pleasure, but ultimately may be far more instrumental in automated AI workflows.
… to save you time
Table summaries are so much bigger than this. tack on these benefits:
To enable data aggregations that are suitable for AI inferencing
To distill the data needed for specific AI queries
To create classes of data distillations that when combined, allow for far greater inferencing possibilities.
What about Embeddings?
Well, this is where we need to go. Imagine an AI query that intimates the need for dynamic integrations from three different data sets. But the query uses words and terms that while similar, are not the same as the data sets such aggregations need to cascade from.
Embeddings offer a way to “set the table” both literally and metaphorically. We need to be able to infer meaning and intent concerning data resources before we can aggregate the right data resources and complete the AI request.
So excited by these! Thank you!
I spent some time learning how to concatenate and then more recently how to use format. Love how this makes it so much easier and faster.
I’ve spent a ton of time aggregating data and getting frustrated stratrated, so excited to explore the new feature!
I’ve started to use AI and excited to learn more, I think there is a lot of potential!
Thank you for the ongoing development of Coda! Coda
Hi, @John_Pia! Unfortunately, we currently don’t support viewing canvas and compose columns in the mobile table view. It’s on our radar as an area of improvement. In the meantime, would making a text column that references the compose column work for you?