scuttleblurb

Share this post

User's avatar
scuttleblurb
[APPN, PEGA, SMAR] No/low code evolution

[APPN, PEGA, SMAR] No/low code evolution

scuttleblurb's avatar
scuttleblurb
Jun 11, 2020
∙ Paid

Share this post

User's avatar
scuttleblurb
[APPN, PEGA, SMAR] No/low code evolution
Share

There are some smart people who think that our brains are wired to process the world hierarchically12.  For instance, neurons in one part of the visual cortex fire when they detect lines tilted at particular angles, sending those patterns to a “higher” layer where certain other neurons are attuned to the detection of particular letters, and so on with words, then clauses, then sentences. 

We dynamically create and alter hierarchies of abstraction, most times oblivious that we are doing so. We don’t cogitate on the millions of muscle movements required for even the simplest of tasks….those details are unconsciously aggregated into high level actions like drinking coffee and brushing one’s teeth, which in turn build up to the even higher level action of getting ready for work, and so on (Stuart Russell believes that a necessary though perhaps insufficient condition for machines to attain general human level intelligence is this ability to dynamically create and scuttle up and down these hierarchies).

Computer programming, like any other language, has evolved to leverage out inherent affinity to conceptual hierarchy. Primitives like syntax and vocabulary are used to construct sub-modules that build up to programs.

While heralded as a new movement, no-code is really the continuation of a decades long process where fine-grained details are implemented behind higher level abstractions that are easier for humans to work with, from the dot/dash binary of Morse code; to the 0s and 1s of machine code; to higher level languages like Java; to Javascript libraries and frameworks; to APIs.   Fifty years ago, financial models were programmed in Fortran by developers; today, they are built in Excel by analysts.  Twenty years ago, every enterprise had its own server racks and an IT staff with specialized technical skills that configured hard drives.  Today, that complexity is hidden behind public cloud APIs. The instructions a computer receives need not take the form of text strings punched in using a keyboard. They can also be blocks and arrows wired together using a mouse.

To offer a personal example of how far we’ve come in this journey, I used to manually enter the scuttleblurb subscriptions processed by WooCommerce, my blog’s e-commerce engine, into an Excel spreadsheet.  While WooCommerce and Excel have APIs that allow the two applications to communicate, tying them together requires some competence in JavaScript and maybe an understanding of how to reuse React components, which I didn’t feel inspired enough to learn.  But then I discovered Zapier, a no-code back-end system that abstracts away all those details, gluing various APIs together without any code, like this:

That’s an actual screen shot of the integration process, which took 10 minutes to complete.

Zapier is just one in a large scrum of no-code companies like Bubble, Airtable, and Webflow that have gained prominence over the last few years.  The latter 3 products are like website builders on steroids, more oriented towards the front-end.  Building them was an arduous and technically demanding challenge because they had to be robust enough to scale and generic enough to accommodate data schemas that couldn’t be predicted in advance.  Bubble’s founders were hazarding guesses on the kinds of queries that had to be optimized and building indices accordingly on the fly3.  Airtable couldn’t find an appropriately scalable off-the-shelf database backend – MySQL can’t handle long strings of dependencies where a change to one cell propagates to many others – so it literally built is own. Rather than rely on a standard Java framework like React, the company also developed its own in-house front-end framework4 to better handle idiosyncratic optimizations.

It took years for Airtable, Webflow, and Bubble to build something that venture capitalists would take seriously.  “What you see is what you get” visual HTML editors from the ‘90s like Microsoft’s FrontPage and Netscape Composer flamed out and besides, in 2012-2013, when these companies were founded, VCs weren’t really interested in a funding a web-based product when it seemed everything was about mobile.   

But there have been major improvements in infrastructure and technology since version 1.0 no-code.  Airtable and Bubble run on AWS and leverage frontend development frameworks like React and distributed databases like MongoDB.  They are surrounded by APIs like Stripe and Twilio that obviate the need to build essential and commonly used functions from scratch.  The V8 browser engine, developed by Google for its Chrome browser and open sourced in 2008, directly translated JS code into machine language (bypassing the creation of intermediate low-level bytecode), dramatically improving JS execution speed and browser performance. In 2011, the release of WebGL, which made it possible to natively render 3D graphics in a web browser, gave rise to front-end design platforms like Webflow and Figma.

Prior to these improvements, visual programming platforms were limited to the enterprise.  They were heavy-duty products with unintuitive user interfaces that required significant training, sold top down by legacy vendors like Pega Systems and Appian, with none of the bottom-up virality enjoyed by modern no-code products.  So while no-code programming has technically been around for decades, only in the last few years, with all the modern foundational infrastructure pieces in place, did it really become democratized and attract VC interest.

Many hard-core coders testily balk at the notion that visual drag-and-drop methods can produce advanced web apps.  And yea, sure, the most sophisticated stuff still requires arduously honed coding skills. 

But first, these platforms have come a long way.  In Bubble, you can supposedly build a JV version of AirBnB within a week, whereas it might take up to 6 months with code.  Dividend Finance, which originates loans for residential solar panels, runs $1bn of lending volume a month on Bubble.  Using no-code tools for its back-end, Lambda School, an 9-month online coding camp, built a solution robust enough to host 3k concurrent students and attract series B venture capital.

And second, evaluating the viability of no-code based on how it performs relative to traditional coding sort of misses the point.    

One of the truisms about software is that it is ~infinitely scalable, with heavy upfront development costs succeeded by ~zero marginal cost delivery.  Whether you’re talking about SaaS or packaged software, scalability depends on broad adoption, which means catering to the average, which means creating a standard solution whose core functions are essentially equivalent across users.  But there is an unfathomably vast array of idiosyncratic workflows and use cases and preferred interfaces specific to startups and enterprises and teams that are too niche to support.  

A useful analogy is gaming, which has transitioned from a state where developers create landscapes, narratives, and characters that appeal to broad audiences; to one where players express agency in terms of the characters they choose and the worlds they navigate; to now, as discussed in this write-up from Matthew Ball, a place where players manipulate visual primitives on platforms like Roblox and Minecraft to create custom digital worlds and maneuver through those worlds as themselves. 

Likewise, no-code pushes programming to ever higher planes of abstraction, allowing tech-savvy but non-technical users to create software that addresses all sorts of granular cases.  With Weblow, domain experts with no coding experience have created marketplaces for surfboards and modular homes.  With Airtable, a goodwill store might upload the pictures of thousands of donated goods and use Google’s Cloud Vision API to label them.  With Parabola, business users can build an application form that, when filled out, triggers a workflow in which the HR team is emailed details about the applicant and the uploaded resume is automatically transferred to a shared Dropbox folder; or a sales team can collect leads from Typeform and Mailchimp, filter for certain characteristics, and clean the two datasets before merging them together and exporting the consolidated list into salesforce.  Bubble lets you create very similar workflows.

[the above is a drag-and-drop automated workflow created in Parabola]

It can get a lot more complicated than this.  A customer interaction might involve dozens of steps that pull data from many siloed systems.  There are millions of edge cases and permutations specific to an organization or team that standard off-the-shelf apps can’t effectively address. 

We seem to be in this sweet spot where the technology that no-code leverages is solid enough and the use cases modest enough (teams aren’t using no-code tools to analyze terabytes of real time data) that the performance trade-off vs. coding is worth it, particularly for non-numeric use cases currently handled by Excel (I once heard a stat that 60% of Excel spreadsheets don’t contain a formula!).  You might think of no-code tools as occupying a vast, albeit nebulous, middle ground: it’s less performant than coding but more efficient to use; more performant than Excel (for certain use cases) but somewhat harder to learn.     

Excel excels in financial modeling – the ability to change one cell and immediately see changes in all the cell’s dependencies is amazing – but was never really designed for tasks that teams are now using it for, like keeping track of internal projects.

The latter is better handled by something like Airtable, a relational database masquerading as a spreadsheet:

[see this Vimeo page for helpful Airtable demos]

Each row represents a record and each record has fields (“columns” in Excel vernacular) that store different kinds of content, including dates, images, pdf’s, audio files, labels from drop-down menus and links to records in other tables.  Clicking on a link – the name of a person assigned to a certain project, for instance – will generate a pop-up window containing all the information about that person stored on another table.  By setting up a third-party integration with Gmail through Zapier, an email notification can be sent to certain team members whenever a new record is added to the table.  Behaviorally, you are manipulating the database as you would a spreadsheet but functionally, Airtable can create relational structures between different tables, which spreadsheets can’t do.  For instance, from an Airtable containing a list of stores, you might pull a list of inventory acquired on a certain date from another Airtable containing information on all the company’s SKUs.  If you edit information in one table, those changes populate in the other table. 

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2025 scuttleblurb
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture

Share