Hey Coda Makers! We’re excited to share some news: Superhuman has agreed to acquire Rows.
First things first: nothing changes about the product you already use and love. Coda, as your one workspace for docs, tables, and workflows, stays exactly the same. What does change is that this acquisition will make Coda even more powerful.
The Rows team has spent years making it easy for people to work with business data; building integrations, helping teams make sense of information, and creating AI tools that turn data into answers. Now, they’re bringing all that expertise to Coda. What that means for you: we can do even more for data-driven teams, and we’re building toward a future where your data, communication, and documents work together seamlessly.
We’re really excited about what this opens up, and we’ll keep you in the loop as things take shape.
If you want the full story, check out the blog or Shishir’s LinkedIn post.
13 Likes
I really admire Rows as a product and as a team! I’m sure we’ll soon be able to notice many improvements in Coda!!
3 Likes
I am super excited by this news.
rows (previously known as DashDash) has brought several major innovations to the world of no-code automations (using an extension of the EXCEL spreadsheet paradigm).
- Browser-based spreadsheet engine - turns sheets into collaborative web apps
- Data integrations GET() and POST() functions to access data via any HTTP REST API
- Machine Learning (ML) neural network AI features (way back in 2018) for finding patterns in numeric data; CLASSIFY() & PREDICT()
- LLM AI features (mid 2023) for text processing and ‘formulas in English’; AI() & AI_TRANSLATE()
So I am hoping to see these features migrated into the Coda Tables and Coda Formulas engines over the next few months.
That would allow us to write CFL formulas that can perform GET() and POST() API fetches to external data sources. The question is; will this be available without the need to write a Pack for each integration?
And the other major feature I am hoping to see in CFL is the ability to execute TensorFlow ML functions like CLASSIFY() and PREDICT() on large sets of numeric data, where the system will build and train neural networks on-the-fly and use them to detect patterns in the numeric data.
(I tried to build my own TensorFlow pack but failed; the training times exceeded the 60-second time-out limit for packs.)
Right now, LLM-based AI functions are very bad with numeric data - they can only process text-based information with any precision.
Disclaimer:
I am not party to any internal discussions or planning inside Superhuman.
So all this is based on my knowledge of the rows & Coda products and my own speculation (and wishful thinking).
I await developments with bated breath!
➤𝖒𝖆𝖝
6 Likes