We are building a modern spreadsheet and experimenting with different ideas around improving robustness of spreadsheets, making it easier to tell stories with data, and helping both casual and power users to speed up their work with AI.
On a technical side we wanted to make an online spreadsheet that exceeds the speed of a desktop version of Excel. We are doing it because I had to often juggle between GSheets and Excel for different use cases in my previous company. And also because both myself and our team love performance optimization.
In the blog post above we are sharing early benchmarks. We will write a separate article about our tech stack, but in short - we are running all calculations on the client-side using multithreaded WebAssembly (C++) & CRDT for collaboration. We also wrote our own optimized Canvas-based spreadsheet renderer.
Running calculations locally allows us to minimize cost to serve a user, to eventually implement e2e encryption, and to provide an offline mode. We will also ship a native desktop app next month to get around 4GB browser WASM limitation and to get an extra 30% speed boost.
We are still in early stages, but we'd love to hear your feedback about the product.
Thank you for the question! We haven’t found one silver bullet yet. We instead review each feature and see if we can do it better or differently. Our goal is to provide a better alternative to Google Sheets / Numbers because we feel that moving people away from the Microsoft’s ecosystem is a much harder task.
Some examples:
- Storytelling with data
- Charts: Many users told us that our chart creation process is easier than GSheets and Excel. We are also making it easier to add annotations to charts (CAGRs, Value lines etc.)
- Visual board / Dashboard: Easy to make a numbers-heavy presentation right inside the app instead of copy-pasting charts into Google Slides and spending hours on alignment of elements.
- Speed (vs GSheets and Numbers):
- I often saw how people would start building their financial models in Google Sheets and would then switch to Excel once the model grows. We hope that some users would give us a try instead of switching to a desktop Excel.
- Anti-fragility (vs everyone)
- Models become fragile as they grow. We are building a bunch of features such as highlighting cells based on the underlying data or making it easier to find a root cause of an error
- AI:
- Do basic data transformations and data cleaning without in-depth knowledge of formulas
- Better mobile / iPad experience (WIP, not released yet)
The plan is to keep building until it feels much better and some people would consider switching despite their current tool being good enough :)
This not a complaint ... rather it is based on observations over the last 25+ years :)
No one cares about you doing a "feature" "better or differently"!
Let me use a real-world example: no one cares if your hammer is "better" or "different" than all other hammers for the last uncounted millennia - they care that they can hit a nail accurately and sink it with a reasonable amount of effort
>Speed
This is NOT a problem to solve! [Practically] No One is ever recalculating their whole sheet! And even if they happen to be doing so, the %% of users for whom this is even close to mattering is vanishingly small (in 25+ years, I have met two people ever for whom this might matter - my dad was one (a true Excel Power User from the late 80s on to his retirement a few years ago ... his spreadsheets could bog-down 16-core systems with 32G RAM because of their size, reliance on network resources, and massive number of formelae present))
>Anti-fragility
...what does this even mean? "Models become fragile as they grow" ... and? Who out there is really bumping up against fragility of their underlying models?
>AI: - Do basic data transformations and data cleaning without in-depth knowledge of formulas
Explicate, please! Define "data cleaning". Define "data transformations". It is rarely a lack of "in-depth knowledge of formulas" that affect data reporting ... it is far more often a lack of in-depth knowledge of the data that leads to issues in reporting
>Better mobile / iPad experience (WIP, not released yet)
Ok - this might matter ... Excel, Number, Google Sheets, etc are far better when navigable via mouse+keyboard vs touch interface :)
...but! If you have a keyboard & mouse for your tablet, this is 100% moot :)
>The plan is to keep building until it feels much better and some people would consider switching despite their current tool being good enough :)
Aye! There's the rub!
As has been stated myriad times in the past, "'good enough' - most often - is 'good enough'".
I wish you the best in your endeavors ... but make sure you are NOT trying to merely 'augment the status quo' - build something new/different! Just like VisiCalc was groundbreaking in the late 70s because it was brand-flipping-new and different ... you need to be THAT!
This, I think, is a reason AirTable (no affiliation, just have benefitted from them a few times) has been [relatively] successful - they are different from merely "Excel in the cloud"
Google Sheets made waves by being multipli-live-editable in the cloud ... and then LibreOffice, O365, Apple Numbers, etc added the same thing
Now? The only differentiator on the Google productivity suite is 'seamless' integration with Google accounts
You need to find your Differentiator - maybe you have it, but it is sitting on the sidelines shouting to the coach, "put me in, I can play!"
Maybe you have not yet recruited that player to the team :)
fwiw ... if you would like to take this convo [quasi] offline, I would be happy to engage via email
should be in my profile - if NOT, warren [at] antipaucity [dot] com
Appreciate your points but I personally disagree, I definitely need speed — closing my startups month end books took days because file was too big and slow to process online, and sharing offline files between teammates became confusing. So have been watching this companies progress and excited to see the speed, and other, results.
>closing my startups month end books took days because file was too big and slow to process online
Whahuh?
Maybe what you say is correct ... but it sounds like you were trying to keep all your 'books' in a single spreadsheet for all time ... with maybe a lot of [poorly-written] formulae or macros running?
-----
Also ... this is why 'real' accounting tools like Quickbooks exist - a spreadsheet is not always the 'right answer' :)
I must say that I am using these spreadsheets from time to time for personal notes and stuff, and they feel pretty smooth and fast, way better than google spreadsheets. In terms of basic user experience.
Btw my own small documents did not work in Safari on iPad, which I would love as iPad user with the keyboard, on iPhone read-only of this article works fine.
8 comments
[ 3.0 ms ] story [ 28.9 ms ] threadWe are building a modern spreadsheet and experimenting with different ideas around improving robustness of spreadsheets, making it easier to tell stories with data, and helping both casual and power users to speed up their work with AI.
On a technical side we wanted to make an online spreadsheet that exceeds the speed of a desktop version of Excel. We are doing it because I had to often juggle between GSheets and Excel for different use cases in my previous company. And also because both myself and our team love performance optimization.
In the blog post above we are sharing early benchmarks. We will write a separate article about our tech stack, but in short - we are running all calculations on the client-side using multithreaded WebAssembly (C++) & CRDT for collaboration. We also wrote our own optimized Canvas-based spreadsheet renderer.
Running calculations locally allows us to minimize cost to serve a user, to eventually implement e2e encryption, and to provide an offline mode. We will also ship a native desktop app next month to get around 4GB browser WASM limitation and to get an extra 30% speed boost.
We are still in early stages, but we'd love to hear your feedback about the product.
Not because I do not admire your efforts
But because I am NOT going to go learn yet another tool for funzies when I already have half a dozen that already Just Work™ :)
Some examples: - Storytelling with data - Charts: Many users told us that our chart creation process is easier than GSheets and Excel. We are also making it easier to add annotations to charts (CAGRs, Value lines etc.) - Visual board / Dashboard: Easy to make a numbers-heavy presentation right inside the app instead of copy-pasting charts into Google Slides and spending hours on alignment of elements.
- Speed (vs GSheets and Numbers): - I often saw how people would start building their financial models in Google Sheets and would then switch to Excel once the model grows. We hope that some users would give us a try instead of switching to a desktop Excel.
- Anti-fragility (vs everyone) - Models become fragile as they grow. We are building a bunch of features such as highlighting cells based on the underlying data or making it easier to find a root cause of an error
- AI: - Do basic data transformations and data cleaning without in-depth knowledge of formulas
- Better mobile / iPad experience (WIP, not released yet)
The plan is to keep building until it feels much better and some people would consider switching despite their current tool being good enough :)
No one cares about you doing a "feature" "better or differently"!
Let me use a real-world example: no one cares if your hammer is "better" or "different" than all other hammers for the last uncounted millennia - they care that they can hit a nail accurately and sink it with a reasonable amount of effort
>Speed
This is NOT a problem to solve! [Practically] No One is ever recalculating their whole sheet! And even if they happen to be doing so, the %% of users for whom this is even close to mattering is vanishingly small (in 25+ years, I have met two people ever for whom this might matter - my dad was one (a true Excel Power User from the late 80s on to his retirement a few years ago ... his spreadsheets could bog-down 16-core systems with 32G RAM because of their size, reliance on network resources, and massive number of formelae present))
>Anti-fragility
...what does this even mean? "Models become fragile as they grow" ... and? Who out there is really bumping up against fragility of their underlying models?
>AI: - Do basic data transformations and data cleaning without in-depth knowledge of formulas
Explicate, please! Define "data cleaning". Define "data transformations". It is rarely a lack of "in-depth knowledge of formulas" that affect data reporting ... it is far more often a lack of in-depth knowledge of the data that leads to issues in reporting
>Better mobile / iPad experience (WIP, not released yet)
Ok - this might matter ... Excel, Number, Google Sheets, etc are far better when navigable via mouse+keyboard vs touch interface :)
...but! If you have a keyboard & mouse for your tablet, this is 100% moot :)
>The plan is to keep building until it feels much better and some people would consider switching despite their current tool being good enough :)
Aye! There's the rub!
As has been stated myriad times in the past, "'good enough' - most often - is 'good enough'".
I wish you the best in your endeavors ... but make sure you are NOT trying to merely 'augment the status quo' - build something new/different! Just like VisiCalc was groundbreaking in the late 70s because it was brand-flipping-new and different ... you need to be THAT!
This, I think, is a reason AirTable (no affiliation, just have benefitted from them a few times) has been [relatively] successful - they are different from merely "Excel in the cloud"
Google Sheets made waves by being multipli-live-editable in the cloud ... and then LibreOffice, O365, Apple Numbers, etc added the same thing
Now? The only differentiator on the Google productivity suite is 'seamless' integration with Google accounts
You need to find your Differentiator - maybe you have it, but it is sitting on the sidelines shouting to the coach, "put me in, I can play!"
Maybe you have not yet recruited that player to the team :)
fwiw ... if you would like to take this convo [quasi] offline, I would be happy to engage via email should be in my profile - if NOT, warren [at] antipaucity [dot] com
Whahuh?
Maybe what you say is correct ... but it sounds like you were trying to keep all your 'books' in a single spreadsheet for all time ... with maybe a lot of [poorly-written] formulae or macros running?
-----
Also ... this is why 'real' accounting tools like Quickbooks exist - a spreadsheet is not always the 'right answer' :)
Btw my own small documents did not work in Safari on iPad, which I would love as iPad user with the keyboard, on iPhone read-only of this article works fine.