How Low Can You Go? A retrospective
Over four years ago now, in mid-2021, I wrote a series of Refractis “Angles” about the increasing adoption of Low Code tools. Alongside it, I published a directory of 54 tools which spanned Code, Data and Integration tooling.
The tools were:
| Category | Tools | Count |
|---|---|---|
| Integration Tools | Apigee, Appian, Automate, Boomi, IFTTT, Integromat (now Make), Jitterbit, n8n, Salesforce Mulesoft, Workato, Zapier | 12 |
| Data Tools | Airtable, Caspio, Knack | 3 |
| Development Tools | Adalo, Appenate, Appmysite, Appsheet, Backendless, Betty Blocks, Bravo Studio, Bubble, Buildbase, Candu, Flowfinity, Genexus, Glide, lansa, Mendix, MS Power Apps, Nutshell Apps, OutSystems, Pandasuite, Quickbase, Rocket LegalSuite, Salesforce Lightning,Soti snap, Swing2app, Thunkable, Twixl, Unqork, Voiceflow Webflow | 29 |
| Other | Airslate, Appdome, Bryter, Carrd, Kintone, Quixy, ServiceNow Platform, Spreadsheetweb, Stencyl, Tines | 10 |
The headline findings: 53 of the original 54 tools still exist. AI Adoption is almost ubiquitous, and many of the adoption guidelines are still relevant today.
AI has changed the landscape and is almost ubiquitous, although I haven’t investigated whether this is genuinely useful or marketing spin. Of course, this isn’t the only market which has been transformed in ways that could not have been anticipated in 2021.
I can't claim amazing prescience at this survival rate, I think the market generally has been good for all low-code tools, and the ones I selected were already leaders in their space.
There was a mix of well-funded start-ups (Airtable has raised $1.4bn in funding) Gartner leaders and large tech parent involvement (Salesforce, Google) all of which will have helped ensure survival.
In my original conclusion I noted there were lots of low code tools and platforms out there, that they were evolving quickly, and that new ones were joining the market all the time. I predicted consolidation, changes in commercial models, and even failures. I advised users to choose tools or ecosystems that suited their requirements at the time, to keep watching the feature announcements and the wider market, and to be prepared to switch.
I thought it would be interesting to revisit the articles, review my original list, see which of the tools had thrived, and how they had adapted to the changes that AI has brought to the space.
Here's my review and update on each of those articles in turn.
Part 1: Integration — 'Beating the System Integrators'
Back in 2021 I highlighted Zapier, Workato, Automate.io, and Integromat as integration leaders. I warned about Shadow IT risks, recommended corporate standards with appointed 'champions', and cautioned that each of these tools required more technical skill than the marketing suggested—they were/are 'low code' not 'no code'.
My closing prediction: implemented correctly, Integration tools would reduce the need for System Integrator projects stitching applications together.
Updates:
Zapier has gone from strength to strength and positions the tool as 'the most connected AI orchestration platform' with 8,000+ app integrations, AI Workflows, AI Agents, and AI Copilot.
Workato: Remains strong with 'AI by Workato', Agent Studio, and EU-compliant agentic AI services. Enterprise focus validated.
Integromat: Acquired by Celonis and rebranded as Make—largely with little impact to workflows, although migration was required for Webhook URLs ahead of a well-publicised deadline.
Automate.io: Acquired by Notion shortly after the article was published, and was shutdown, leaving customers to migrate to alternative platforms.
Verdict:
The Shadow IT warning and 'champions' recommendation remain highly relevant.
The prediction about failures and consolidation proved accurate (Automate and Integromat).
The caution about technical skills required has been somewhat mitigated by AI assistants that genuinely lower the barrier to entry.
Part 2: Data Platforms — 'I've Become a Big Fan'
In 2021 I declared myself 'a big fan' of Airtable specifically, noting its Unicorn status ($1bn+ valuation). Back then I had started using it for migration tracking (1,000+ migrations) at my client and for financial tracking at Refractis.
I recognised the power of low-code data platforms as replacements for complex spreadsheets where relational data was needed, whilst acknowledging overkill for simpler use cases where spreadsheets sufficed.
Updates:
Airtable exceeded even optimistic expectations. It has developed a particular niche in product management. CEO Howie Liu claims the platform has been 'refounded' as an AI-native platform. A builder tool, Omni, was introduced, and several ring-fenced AI models are available from within the platform. Commercially Airtable has been successful, with 27% YoY growth whilst achieving positive cash flow.
Verdict:
My enthusiasm hasn’t abated (and I think has been valuable). Airtable has evolved from 'spreadsheet replacement' to 'AI-native app platform'. The prediction that data platforms had 'even more potential to disrupt' has been validated—though AI was the catalyst, not just relational data capabilities.
Part 3: Development — 'Making Developers More Efficient'
Four years ago I cited Gartner's $14bn market prediction, highlighted Appian, OutSystems, and Genexus as targeting existing coders rather than citizen developers. My key insight: low-code makes developers more efficient, rather than replacing them.
I mentioned Microsoft PowerApps was likely to become 'the default choice for citizen developers in enterprises running O365'. I confessed to not being good at coding myself, and joked that 'on name alone it might have to be Betty Blocks'.
Updates:
Gartner’s market predictions were dwarfed: The market reached $30.1B in 2024 and is projected to hit $101.7B by 2030—far exceeding the 2021 forecasts.
Appian & OutSystems: Both remain Gartner Magic Quadrant Leaders for the 9th consecutive year. Appian has 50,000+ AI agents deployed
Microsoft Power Apps: Has indeed become an enterprise default, now with Copilot AI integration deeply embedded
Betty Blocks: Still active with strong AI features including generative UI design and 'Betty Genius' AI assistant—my selection turned out to be a solid platform rather than just an appealing name
Bubble: Emerged as a standout with 48% YoY revenue growth, 1M+ customers, and an AI Agent for automated app building
Verdict:
The 'developer efficiency' thesis has been superseded by AI. The barrier to entry has genuinely lowered—AI copilots now assist with the code that 2021-era platforms still required. Non-coders can now achieve results that would have required scripting knowledge four years ago.
Part 4: Conclusions — Testing the Guidelines
Originally, my biggest takeaway was that 'low code platforms overstate their accessibility to ordinary business users'you still needed technical skill and time. I warned to 'expect consolidation, changes in commercial models, and some failures' and recommended being 'prepared to switch'. I published seven guidelines for success.
Updates
| 2021 Guideline | Validity | Update |
|---|---|---|
| Establish requirements and design before building | Still valid; even if this is just to craft your prompt | AI makes building faster but poor requirements still doom projects |
| Document and maintain what you build | Still valid |
With increasing number of tools, shadow IT is an even greater threat. AI Can be used to document what is deployed. |
| You might need to write some code | Less important | AI copilots can now likely write the code for you, or help debug if you get stuck. Expect further improvements as models evolve |
| Empower citizen developers, nominate champions | Still valid |
It still makes sense to establish platform standards and a COE. Gartner now calls these 'fusion teams' |
| Select powerful engine for existing developers | Still valid | Appian, OutSystems still lead here |
| Track spend and outcomes vs traditional methods | Still valid | Tool proliferation, subscription costs and new features can drive up spend |
| Be prepared to re-platform if demand outgrows tooling | Still valid | Confirm requirements (and spend) regularly. Be prepared to pivot based if demand or supply-side change |
Verdict:
Six of the original seven guidelines remain directly applicable. Only the 'you might need code' guideline has been materially changed—AI copilots are evolving to handle much of what previously required scripting. The warning about consolidation (Integromat) and failures (Automate.io) proved accurate, but at a much smaller scale than I expected.
The original list is here:
https://airtable.com/app4E1L8U8gtkRn48/shr8EeSAlHHTXSXGH
I've added a new field to the original directory listing, and for full disclosure, at the end of last year I used Chat GPT to populate it. Using an “agent field” in Airtable, I asked it to verify the vendor was still operating, the URL was still available, and to provide insight on commercial or competitive changes, AI marketing claims and capabilities.
Amazingly, 53 of the 54 tools still exist.
The original series was published in mid-2021, eighteen months before the emergence of ChatGPT. AI has fundamentally transformed the low-code landscape in ways the 2021 analysis could not anticipate, allowing prompting to replace scripting, reducing the barrier to entry.
AI Capabilities have become another differentiator, below is a quick assessment of which of the original vendors have emerged as leaders:
Conclusion:
The original 'How Low Can You Go?' series holds up remarkably well. The directory selection criteria proved sound, my specific platform enthusiasm (Airtable) was vindicated, and most guidelines remain applicable. The main evolution is that AI has made low-code more accessible than anyone dared hope back in 2021, while simultaneously raising the stakes for platforms that fail to embrace it.
For enterprise use, requirements definition, maintainability, championing and monitoring spend and the wider market all remain relevant, as they do with any software selection.
Finally:
To wrap up, here at Refractis we use Airtable, Make, Activepieces and Claude to help integrate and give insights into our back-office systems. Admittedly, our requirements and core toolset are both relatively simple, but we're building increasingly complex and intelligent workflows to support our business as it grows.
One thing we have noticed is subscription costs to ratchet up and vendor lock-in becomes real, quickly. It may even be time to look at alternatives, Gist looks interesting!