Testers wanted: Use words to create expressions

Hey everyone! We are early in the design and development phase of a new feature where app creators can use their own words to create expressions. It is currently enabled for writing slice filter expressions and supports a small handful of expressions. We plan to expand the supported locations in the editor and expressions as we develop the feature.

If you use slice filter expressions and are interested, please reply to this thread to opt in. Our team will follow up separately to ensure you meet the criteria for this particular feature. Not everyone who opts in will be the right fit, but there will be plenty of future opportunities to participate. All levels of familiarity with expressions are welcome to opt in!


Why add this feature?
Writing AppSheet expressions (also called formulas) is hard and time-consuming, especially for a no-code user. The goal of this feature is to make writing expressions easier by enabling users to describe them with words, instead of typing full AppSheet expressions. Less time crafting formulas means more time to add functionality to apps and use them.

How will this affect my existing user experience?
There should be minimal interruption in your current user experience. Creating a new expression will show suggested ones first, but if you find none of them relevant, you will be able to bypass the suggested formulas and use the Expression Assistant instead.


I’ll give it a whirl.
My testing account:
ID: 546920

So natural language reading to give me a formula? Please let me turn this off when it eventually becomes a full feature.
Can you turn this on for single accounts in a Enterprise Team?


So natural language reading to give me a formula? Please let me turn this off when it eventually becomes a full feature.

Point taken!

Can you turn this on for single accounts in a Enterprise Team?

Yes, it’s like a typical rollout where it’s enabled by account ID.

That’s the fastest preemptive opt out i have ever seen, majestic.

I wonder how it would interact with the qrew tools autocomplete function that was added recently.



Good question – the suggestions are currently outside of the Expression Assistant, so I don’t anticipate it interrupting the qrew tools features.


In this contexts, Language, Words means “English” only, i reckon.

i just want to have options to turn them off…


Yep, please ensure a way I can keep this feature out of the way. So many of the features designed to help the less-skilled app creators make my experience more difficult.


Out of my humble and personal opinion, based on my best-guesstimation in terms of how this new feature may work.
I simply assumed we enter the natural languages like we ask Siri or Google, “Hey Appsheet, I wish to get the number of the count for Col1 which is sitting on Table-A. Make sure I dont need the duplicates, so you need to eliminate such and just count distinct count”. Then appsheet may say, “okey, here is the result”.

If it is true, can we verify the appsheet is returning the result of what we need? I dont trust the result, unless I see the proper appsheet expressions.

Yes, it is possible Appsheet the convert such “Words” onto the proper expression. Then we can assess it, if it is okey or not. But for me, rather than typing the natural language, it is quite faster and easier to write the appsheet expression from the begnning.
Even for those who are new to the Appsheet, how can they interpret and verify the appsheet expression which was converted from the natural language? As far as they dont know much about the appsheet expression then they may select a choice to do from natural language to buid the app. But for me it is too dangerous and risky. A

Not sure how the new feature is going to behaive, but for me, it looks not really useful to be honest.

We have bunch of required features for years, I wish Appsheet team to priotarize those feature requests which have been posted by the community member here for years and years. I never see any request from anyone in this commuinty to introduce such as “natural lauguage/words” expressoin feature before. yes it is possible some may requested before, but it should not be supported by the majority of the members.

I have a grave concerns that our voice posted here for the new features are not really taken by the appsheet, but appsheet (now Google) started to take and follow the path what they want, leaving the users voice and request behind… I saw such tihngs for services by Google.
I m really worry about now Appsheet is now really Googlinized in the prefered way.


I agree.

I agree.

I agree.


Agreed on all points. Especially about prioritizing long-standing feature requests.


Triple agree. New cool features are nice but only when they’re coming alongside features that people actually express wanting. I know Appsheet has their own vision for what Appsheet should be but feels like why bother with requesting a feature. I mean if you see things with 40+votes that’s going on its second year of crickets; what’s the point? Even just acknowledgement of them would be good. Not saying implement ever FR with more than 30 votes but maybe respond to them.

Examples of highly desired features with no interaction:

76 votes, not a word on it.

55 votes, not a word on it.

Both of these request have been out there for more than 365 days without a comment from Appsheet.

42 votes, even has a pole on it as well that shows that nearly every recognizable icon from the forum has voted on this.

There’s even a whole bunch of small usability request as well like editor dark mode, some additional formulas for already existing formulas, adding existing features to other parts that just don’t have them, and so many more.



@Steve @GreenFlux @Austin_Lambeth @Rafael_ANEIC-PY

Don’t forget that we have always had the “natural language version” of our expressions available. It’s that blurb of text just below expression where it tells us that our expression is good (and where some of us read to verify our expression is doing what we think it is doing)


I feel like that is binned version of expression to text not the other way around?


This natural language version of an expression isn’t what people would be entering into the suggestion box to get out expressions.

The discussion is very exciting and I agree with all the reactions. Especially to @tsuji_koichi reactions.

hey folks, thanks and a few comments:

  1. One clear takeaway for us: provide more transparency into our feature work, including feedback in the Feature Requests section of the community. This is 100% fair criticism. We recognize it too. Some of you like @GreenFlux and @tsuji_koichi spent the time to share thoughts with us in detail. We’re starting to do some positive things in this direction. We now have daily change logs. Coming soon is some forward looking change projection looking out a few weeks at a time to tell you what’s in the pipeline. I scheduled one webinar in December (and one more this morning for platform partners) on the detailed roadmap for the year. We’ll translate this back into the granular feature requests for more transparency.
  1. Last year, we had to skew most of the resources onto infrastructural stuff. It slowed down the growth in the platform features. Lots of new team members, new environment, everyone working remote, move from Azure to GCP. It was a frustrating year in terms of moving the platform forward with features I know. Still you expect us to do better or at least recover our momentum. I totally agree.

  2. This year, we’ve got a reasonably balanced plan across different things. The fact that we do something for new users doesn’t mean we don’t care about existing users. We have to care about both sets of users. Just like we have to do stuff for small customers and we have to do stuff for big enterprises. I just feel we start failing as a broader group when we end up in an either-or conversation. The truth is that not all engineers are “fungible” — ie. they don’t all have the same skills to work on the same things. Someone working on UI cannot work effectively on running SQL queries faster. There is a balance in the construction of the team and therefore in the work that gets done and how things move forward in parallel.

So I also have a request, especially for the long-term experts on this community. We all know the platform wouldn’t be here without your pushing us and nurturing us at every step. Some things need pushing and some things need nurture. @natalie is doing some really good work to help people who aren’t familiar with formulas and expressions. It may not be useful to an expert and that is ok. It is clearly not meant for an expert. However, let’s still allow the work to benefit from the feedback from people who do see value in the feature. You’re increasingly going to see many threads pre-viewing upcoming features. Some of them will be interesting to you but perhaps not to someone else. There’s always other vehicles to say “but what about feature X that isn’t what we’re talking about here”.

Still, I don’t want to ignore the three specific features mentioned. I’ll touch on them briefly here but also don’t want to hijack the thread into a discussion of these features.
*) Filtering on columns is active work in progress with internal demos already. You will see that roll out soon.
*) We couldn’t make progress on action parameters because we decided to revamp a lot of the workflow rules into a richer process automation mechanism. So that stuff needed to land first (we have to build some things in common across the two of them). It is launching today in preview. By the time we get to the end of Q1 and the full release, we’ll be much closer to where we need on that.
*) Tool tips on action buttons is an instance of the general problem that we haven’t focused enough on desktop/browser UI. This is a big focus for us this year. In fact, it is the single biggest focus for app UI this year.
:]. I know we are all on the same team, and that’s where the passion comes from — strong positive or strong negative. We’ll keep moving it forward.


Understood. I was just pointing out that it’s there.

To reverse-engineer and simplify it would not be difficult - most of the hard translation work is already done.

If you were to look over several of the “worded” expressions, you’ll begin to notice frequent patterns and much of the expression doesn’t really need to be typed - the values could be presented with a dropdown. In fact only static string values would need to keyed in. The rest could be presented in choice lists.

If the UI were built correctly with ease of use in mind, we will be able to build even complex expressions with just a few clicks of the mouse. Should it be the only way? Absolutely not!!

By the way, this is not novel to AppSheet. Many platforms already provide this point-and-click approach to building expressions.

1 Like