Low-Code/No-Code in 2026: What Died, What Survived, and What Developers Should Actually Worry About

Published on BirJob.com · March 2026 · by Ismat

The Presentation That Aged Poorly

In 2021, I sat in a conference room in Baku watching a vendor demo a no-code platform. The sales pitch was breathtaking in its confidence: "Your developers are your bottleneck. With our platform, your business analysts can build production applications in days, not months. You'll need 70% fewer developers by 2025." A few people in the room nodded. One executive whispered to another that this could save them millions.

It's now March 2026. That vendor's platform still exists, but the company using it still has the same number of developers — actually more, because they needed people to maintain the "no-code" applications that became unmaintainable once the original business analyst who built them left. The executive who whispered about savings never mentions it anymore.

The low-code/no-code movement was one of the biggest stories in tech from 2019 to 2023. Billions of dollars in venture capital. Massive acquisitions. Gartner predictions that reshuffled enterprise strategy decks worldwide. Now that we're squarely past the predicted timeline, it's time to do something the hype cycle rarely allows: look at what actually happened.

Not what was predicted. Not what VCs wanted. What actually happened.

The Numbers First

Let's start with the predictions and then match them against reality.

• In 2021, Gartner predicted that by 2025, 70% of new applications developed by organizations would use low-code or no-code technologies, up from less than 25% in 2020.
• Forrester's 2023 report estimated the low-code market at $13.2 billion, projecting it would reach $50 billion by 2028.
• The Grand View Research report valued the global low-code development platform market at $22.9 billion in 2024, growing at 23.1% CAGR.
• McKinsey's developer velocity research found that low-code tools improved delivery speed for internal tools by 40–60%, but had minimal impact on customer-facing product development.
• LinkedIn's 2025 Jobs on the Rise did NOT include any low-code-specific roles in its top growing titles. Software engineer, data engineer, AI/ML engineer — yes. "Citizen developer" — no.

So the market grew. The money was real. But the "70% of apps" prediction? Let's examine that.

Gartner's own 2025 follow-up research acknowledged that while low-code adoption grew significantly, the 70% figure included broad definitions: any use of visual development tools, form builders, workflow automation, or integration platforms counted as "low-code." When you narrow to "full applications replacing traditional development," the real number is closer to 15–25% for internal tools and well under 10% for customer-facing products.

The prediction was technically achievable only because they defined "application" so broadly that a Zapier automation or a Google Form counted. That's not what anyone in the room heard when they saw "70%." What they heard was: most software won't need developers. That didn't happen. It didn't come close.

What Survived (and Is Genuinely Useful)

The low-code/no-code world isn't a graveyard. Some categories found real product-market fit. The key pattern: tools that survived found a specific niche and stopped pretending to be general-purpose.

Internal Tools: Retool, Appsmith, Budibase

Retool might be the single best argument for low-code. It does one thing well: let developers build internal admin panels, dashboards, and CRUD tools quickly by dragging and dropping components that connect to existing databases and APIs. Retool raised $3.2 billion in valuation and actually delivers on its promise because it doesn't try to replace developer tools — it gives developers a faster way to build boring-but-necessary internal tools.

The key insight: Retool's primary users are still developers. They just use it to build admin panels in hours instead of days. The "no-code for business users" dream didn't pan out; the "low-code for developers building internal tools" reality did.

Marketing Sites: Webflow, Framer

Webflow carved out a genuine niche as a visual website builder that produces clean, production-quality HTML/CSS. It's become the standard tool for marketing teams at startups and mid-size companies who need beautiful landing pages without waiting for developer sprints. Framer occupies a similar space with more of a design-tool feel.

Neither of these replaced web development. They replaced a specific category of web development: marketing sites and landing pages. And that's a legitimate, valuable contribution. A marketing team that can ship a new landing page in a day without filing a Jira ticket is genuinely better off.

Workflow Automation: Zapier, Make, n8n

Zapier and Make (formerly Integromat) survived because they do something simple that almost every business needs: connect SaaS tools together. "When a form is submitted in Typeform, create a row in Google Sheets and send a Slack notification." That's not replacing developers — it's eliminating a category of tasks that was never worth a developer's time in the first place.

n8n added the self-hosted, open-source angle and attracted developer-adjacent users who wanted workflow automation without vendor lock-in. All three are thriving in 2026.

Form/Survey/Database Hybrids: Airtable, Notion

Airtable positioned itself as "spreadsheet meets database" and found massive adoption for project management, CRM, inventory tracking, and other structured data use cases. It's not building applications in any traditional sense — it's giving non-technical people a better way to organize information. And it works. Notion similarly became a knowledge management platform that non-developers actually use daily.

Mobile App Prototyping: FlutterFlow, Adalo

FlutterFlow lets you visually build Flutter apps and export actual Dart code. It's found a niche with agencies and startups who need quick MVPs. The "export real code" approach is smart — it gives you an escape hatch when you outgrow the platform.

What Struggled, Stalled, or Died

For every Retool and Webflow, there were platforms that raised huge rounds and then hit reality. Let me be specific.

Outsystems and Mendix: Enterprise Low-Code's Identity Crisis

OutSystems and Mendix (acquired by Siemens) are the two biggest enterprise low-code platforms. Both are still operating and generating revenue. But the dream of "citizen developers building enterprise applications" didn't materialize at scale.

What happened instead: enterprise customers adopted these platforms and then staffed dedicated development teams to use them. The "low-code" platform required specialist knowledge to use effectively. You needed "OutSystems developers" — a job title that exists primarily in consulting firms. The irony is rich: a tool designed to reduce dependence on developers created a new category of developers.

OutSystems' own research showed that most production applications on their platform were built by IT teams, not business users. The citizen developer revolution was, in enterprise, more of a citizen developer suggestion.

The Bubble.io Question

Bubble is fascinating because it tried to be a general-purpose no-code platform for building web applications. And to its credit, people have built real products with Bubble. There are startups that raised funding with Bubble-built MVPs. The platform processes significant traffic.

But Bubble also became the poster child for no-code limitations: performance complaints became a recurring theme, vendor lock-in is total (there's no code export), and scaling to enterprise-level traffic required expensive plans and careful optimization. Multiple startups publicly documented their "migration off Bubble" stories as they grew.

Bubble didn't die. But it settled into a niche: quick MVPs and simple applications that don't need to scale past modest traffic. That's useful! But it's not the "replace developers" vision.

Platforms That Quietly Faded

The casualty list includes smaller players that either shut down, got acqui-hired, or became zombies:

• Unqork — Raised $365 million at a $2 billion valuation. Was supposed to revolutionize enterprise no-code. Underwent significant layoffs in 2023 and 2024. Still operating but far from the vision.
• Betty Blocks — Enterprise no-code from the Netherlands. Struggled to differentiate from OutSystems and Mendix.
• Various AI website builders (2023 era) — A wave of "describe your app and we'll build it" tools appeared. Most produced output that was worse than a WordPress template. The few that survived pivoted to being AI coding assistants (a very different product).

"No-Code Will Replace Developers" — The Post-Mortem of a Failed Prediction

This was the big claim. Let's dissect exactly why it was wrong.

Reason 1: Software complexity is the point. The reason software requires developers is not that we haven't found a good enough visual interface. It's that software solves complex problems, and complex problems require complex logic. You can't drag-and-drop your way to a payment processing system that handles edge cases across 40 countries with different tax laws. You can't no-code a real-time collaborative editor. The complexity lives in the logic, not the syntax.

Reason 2: Maintenance is where software lives. Building version 1 of something is maybe 10% of its total lifetime cost. The other 90% is maintenance: fixing bugs, adding features, handling changing requirements, scaling infrastructure, managing security patches. Low-code tools made the first 10% faster while making the remaining 90% harder. Every team that built something complex in Bubble or OutSystems discovered that maintaining it required just as much skill as building it — but now you also needed platform-specific expertise.

Reason 3: The "last mile" problem. Low-code platforms handle 80% of common use cases well. The remaining 20% — custom business logic, unusual integrations, performance optimization, accessibility requirements — requires dropping into code. And that last 20% is often the part that makes the product valuable. As Jeff Atwood noted, the last 20% of effort produces 80% of the results. Low-code gives you the easy 80% and leaves you stranded at the hard 20%.

Reason 4: Vendor lock-in became obvious. When you build with React, your code is portable. When you build with Bubble or OutSystems, your entire application lives inside their ecosystem. Multiple high-profile companies discovered this the hard way when they outgrew their no-code platform and had to rebuild everything from scratch. There is no "export" button for business logic built in a visual editor.

The "Citizen Developer" Reality

Gartner's thesis was that business users — analysts, operations managers, marketing professionals — would become "citizen developers," building their own applications without IT involvement. Their definition: "A citizen developer is an employee who creates application capabilities for consumption by themselves or others, using tools that are not actively forbidden by IT or business units."

What actually happened:

• Citizen developers do exist — but they build spreadsheets, automations, and simple workflows. They do NOT build production applications. The typical "citizen developer" output is a Zapier workflow, an Airtable base, or a Power Automate flow. Useful? Absolutely. "Application development"? Only by the loosest possible definition.
• Shadow IT exploded. Gartner's own 2023 data showed that shadow IT (technology deployed without IT oversight) drives 60% of tech spending. Citizen developers built things without considering security, compliance, data governance, or integration architecture. IT teams then had to clean up the mess.
• The governance problem. When 50 business users each build their own Airtable-based mini-applications, you get 50 disconnected data silos with no version control, no testing, no documentation, and no disaster recovery. The McKinsey analysis found that organizations benefited from citizen development only when they wrapped it in governance frameworks — which, ironically, required IT teams to build and manage.

The citizen developer revolution is real, but it's not what was advertised. It's not business users replacing developers. It's business users doing more with spreadsheet-adjacent tools, which is something they were already doing before the term was coined. We just gave it a fancier name and a Gartner Magic Quadrant.

Where Low-Code Genuinely Works vs. Where It Fails

After watching this space for five years, the pattern is clear. Low-code is excellent for some things and terrible for others. Here's the honest breakdown:

The pattern: low-code works great for the "ordinary" parts of software and falls apart for the "interesting" parts. And unfortunately, the interesting parts are usually why the software exists in the first place.

Impact on Developer Jobs: The Honest Assessment

This is what most developers actually want to know: should I be worried?

Short answer: No. But with nuance.

The BLS projects 17% growth in software developer employment from 2023 to 2033 — about 327,000 new jobs in the US alone. If low-code were actually replacing developers, that number would be shrinking. It's not.

The Stack Overflow 2024 Developer Survey showed that 62% of developers have used some form of low-code or no-code tool at work — but almost exclusively for internal tools, prototyping, or non-core functionality. None reported that low-code threatened their role.

What low-code did change:

• It eliminated some junior developer work. Tasks like "build an admin panel that shows database records in a table" used to take a junior developer a week. Now it takes a senior developer 2 hours in Retool. This means fewer entry-level "build basic CRUD" jobs, which makes junior developer hiring more competitive. This is a real effect.
• It raised the bar for what custom code needs to do. If a landing page can be built in Webflow, you don't get paid to build landing pages in React. The work that requires developers is increasingly the hard stuff: complex systems, performance-critical applications, novel product features. This is good for skilled developers (higher-value work) and harder for less skilled ones.
• It created new roles adjacent to developers. "Low-code developer," "automation engineer," "Salesforce administrator" — these are real jobs that pay $60,000–$120,000 and exist because of the low-code ecosystem. They didn't replace developer jobs; they filled gaps between business users and developers.

The net effect: low-code raised the floor (less grunt work for developers) while leaving the ceiling intact (complex software still needs skilled engineers). If you're a developer who builds complex systems, you're fine. If you're a developer whose primary value was building things that low-code tools now handle, you need to upskill.

The AI Disruption Angle (Because You're Thinking It)

Here's the twist nobody saw coming: AI coding assistants (GitHub Copilot, Cursor, Claude) may accomplish what low-code couldn't — not by replacing code with visual builders, but by making code dramatically faster to write.

GitHub's own research showed Copilot users completed tasks 55% faster. McKinsey found generative AI increased developer productivity by 20–45% for well-documented tasks.

This matters because the fundamental value proposition of low-code was "writing code is slow and expensive." If AI makes writing code 2–3x faster, the case for visual builders weakens further. Why learn a proprietary visual language when you can describe what you want to an AI and get portable, standard code?

I think this is the biggest threat to the low-code market in 2026. Not developer resistance, not governance challenges — but AI making traditional coding fast enough that low-code's speed advantage disappears.

When to Use Low-Code vs. Custom Code: A Decision Framework

For teams trying to make this decision right now, here's the framework I'd use:

The one-sentence rule: If the primary complexity of your project is in presentation (how it looks) rather than logic (what it does), low-code is probably a great fit. The moment the logic becomes the hard part, you need code.

Salary and Job Market Data

For people considering careers in the low-code space vs. traditional development, here's what the market looks like:

Sources: BLS, Glassdoor, Levels.fyi, PayScale.

Notice: low-code-specific roles consistently pay 15–30% less than equivalent traditional development roles. This makes economic sense — the skills are more accessible, the supply of people who can use visual tools is larger, and the work is generally less complex. The exception is Salesforce, where the ecosystem is so large and enterprise-entrenched that Salesforce developers command strong salaries.

For people in emerging markets like Azerbaijan, Turkey, or Eastern Europe: low-code skills can be a stepping stone to remote work, especially in the Microsoft Power Platform and Salesforce ecosystems. But if you're investing in a long-term career, learning actual programming will always have a higher ceiling.

What I Actually Think

Here's my honest, unhedged take:

1. The "low-code replaces developers" era is over. Nobody serious makes this claim anymore. The surviving platforms have found their niches, and those niches are legitimate. But the grand vision of democratized application development that eliminates the need for professional software engineers? It failed. Not because the tools were bad, but because software engineering is genuinely hard, and visual abstraction doesn't make hard problems easier — it just makes easy problems faster.

2. AI is doing what low-code couldn't. The real threat to "developers as we know them" isn't no-code platforms — it's AI coding assistants that make real developers dramatically more productive. The 10x developer used to be a myth. With AI tools, a skilled developer can genuinely produce 3–5x more output. This makes the gap between "using a developer" and "using a low-code tool" smaller, which paradoxically reduces low-code's value proposition.

3. Every developer should know how to use low-code tools. Not because they'll replace your skills, but because knowing Retool, Webflow, or Zapier makes you more productive. If I can build an admin panel in 2 hours with Retool instead of 2 days with React, that's 14 hours I can spend on the hard, interesting, high-value work. Smart developers use low-code as a tool in their toolkit, not as a replacement for their toolkit.

4. The biggest lesson from the low-code hype cycle is about predictions. When a technology vendor or analyst firm tells you a technology will replace an entire profession, apply extreme skepticism. The track record of such predictions is terrible. Spreadsheets didn't replace accountants. Word processors didn't replace writers. Low-code didn't replace developers. And — hot take — AI won't replace developers either, though it will change what developers do.

5. If you're building BirJob (or any real product), low-code is not an option. I scrape 91 websites, run Playwright for JavaScript-heavy sites, manage a Postgres database with Prisma, serve a Next.js frontend, process payments, and run analytics. There is no low-code platform on Earth that could handle this. And that's true for most products that are interesting enough to build a business around.

Sources

• Gartner — Low-Code Market Forecast (2021)
• Forrester — State of Low-Code Platforms (2023)
• Grand View Research — Low-Code Development Platform Market
• McKinsey — Developer Velocity Research
• BLS — Software Developers, Analysts, and Testers
• Stack Overflow Developer Survey 2024
• GitHub — Copilot Productivity Research
• McKinsey — Generative AI Developer Productivity
• Gartner — Citizen Developer Definition
• Gartner — Shadow IT Spending (2023)
• McKinsey — The Promise and Challenge of Citizen Development
• LinkedIn — Jobs on the Rise 2025
• Glassdoor — Salary Data
• Levels.fyi — Tech Compensation Data

I'm Ismat, and I build BirJob — a job aggregator that scrapes 91 websites daily to bring every job listing in Azerbaijan into one search. No low-code was harmed in the making of this product. Quite a lot of Python, TypeScript, and late nights were.

You might also like

• AI Engineer vs ML Engineer: What Actually Changed and Why It Matters
• Remote Work from Azerbaijan: The Complete Guide for Professionals and Companies
• The Best Free Certifications That Actually Get You Hired in 2026
• Will AI Replace Jobs in Azerbaijan? Here's What the Hiring Data Actually Shows