Today we’re launching two features that solve the same problem from different angles: your data is stuck outside Docs, and getting it in has always been more painful than it should be.
Import your Excel files
Drag an XLSX file into Docs and your tables arrive as real, working tables with formulas included. No more falling back to CSV and losing your work. Row-level formulas, SUMIFs, and VLOOKUPs all map over automatically.
Extract data from PDFs and images in AI Columns
Upload invoices, contracts, receipts, or any document and AI Columns reads the file and pulls structured data directly into a table. No manual entry or copy-paste.
Give them a try, and if you run into anything, let us know here.
A video on how to do the PDF extract would be helpful. I figured out that I needed to upload the file to the AI prompt box but my first inclination was to upload a file to the page and then request AI to convert it to a table.
Here’s a short demo video of how to use AI Columns that we shared a few days ago on social media. Happy to follow on with another, longer video that answers any questions about the feature.
@Andrii_Khymchuk you can use AI Columns to extract data from file, image or classic columns with text, dates, etc. How would you imagine using it to extract data from docs? Would it be by having a column with public URLs and have the AI Column scrape its data?
Nice update - cool to see that formulas will be preserved. When I get some time I’ll do some testing around the following (unless anyone here already has answers?)
How does it behave if a spreadsheet has inconsistent (e.g. cell not column level) formulas.
Any additional details regarding how the excel functions are “translated” to Coda friendly functions?
And separately - any plans to introduce the ability to “overwrite data” based on Row ID (primary key) via this method?
The idea behind my question comes from the way legal documents are stored at the NGO where I worked. The documents themselves are kept on Google Drive, either as DOCX files or Google Docs, while links to them and other metadata are maintained in Coda.
It would be useful if some of these metadata fields could be populated automatically based on the content of the documents, rather than entered manually. For example, the system could generate a short summary of each document, assign it to a relevant category, or extract other useful metadata from its content.
Thanks, Andrii. It’s clear. We’d solve that use case by giving AI Columns Web access (so that they would access those Google Docs). We don’t have that functionality in the immediate roadmap, but I’ll investigate how we could implement it and report back on whether we add it to the short-term roadmap.
