Zapier has over 7 million users and connects more than 7,000 apps. The company is profitable, bootstrapped until a recent growth round, and remains the dominant player in workflow automation.
By all accounts, Zapier has built a sustainable, successful business.
But when I look at how new builders are automating workflows in 2025—connecting tools, moving data, triggering actions—I’m seeing a fundamental shift that questions whether Zapier’s value proposition remains relevant.
Let’s apply the Quicksand Framework.
The Thesis Check
PMF Timeline: Zapier reached product-market fit around 2014-2016, becoming the standard solution for no-code workflow automation between apps.
Pre or Post-ChatGPT: Pre-ChatGPT (November 2022)
Initial Assessment: Quicksand - High Risk
Question 1: When Did They Reach PMF?
Zapier’s breakout period was 2014-2016. The product solved a clear problem: businesses needed to connect different SaaS tools, but building custom integrations required developers and was expensive.
Zapier offered a no-code solution where anyone could create “Zaps”—automated workflows triggered by events in one app that performed actions in another. “When this happens in App A, do this in App B.”
This democratized automation and became essential infrastructure for businesses using multiple SaaS tools.
This means Zapier’s core product philosophy was established 9-11 years before AI agents could handle tool integration autonomously.
Question 2: What Workflow Assumptions Are Baked In?
Zapier was built on these foundational assumptions:
Humans need to design workflows:
- Users manually configure triggers and actions
- Workflow logic requires human specification (”if this, then that”)
- Templates help, but humans still set up and maintain Zaps
Automation requires pre-built integrations:
- Each app needs a Zapier integration to be connectable
- The integration defines what triggers and actions are possible
- More integrations = more value
Workflows are deterministic and rules-based:
- Automation follows explicit if/then logic
- Workflows don’t adapt based on context
- Humans handle exceptions and edge cases
No-code democratizes automation:
- Non-technical users can automate without developers
- Visual workflow builders are more accessible than code
- The value is making automation accessible to everyone
What this assumed about the future: That automation would continue to require humans to explicitly design workflows, and that no-code visual builders would remain the most accessible way to create automations.
Question 3: How Are They Responding to AI?
Zapier has added AI capabilities and launched products that incorporate AI:
What they’ve added:
- Zapier Central (AI-powered business automation)
- AI-generated Zap suggestions
- Natural language Zap creation
- Chatbot builder with AI capabilities
- Integration with AI platforms (OpenAI, Anthropic)
The pattern: Zapier is trying to layer AI on top of their existing automation model. You can now:
- Describe a workflow in natural language and have it generate a Zap
- Use AI to suggest automations based on your tools
- Incorporate AI actions (like GPT calls) into traditional Zaps
But the core model remains: humans design workflows, Zapier executes them deterministically.
What they haven’t done:
- Enable AI agents to autonomously create and modify workflows based on goals
- Move beyond pre-built integrations to AI agents that can interact with any tool
- Shift from “humans design workflows” to “AI agents handle coordination”
- Fundamentally rethink what automation means when AI can reason about what needs to happen
The fundamental tension: Zapier’s value was democratizing automation by removing the need for code. But AI is the new “no-code”—you describe what you want, and AI figures out how to do it. This threatens to make visual workflow builders feel like unnecessary abstraction.
Question 4: Where Are New Builders Starting?
This is where the disruption becomes visible.
Observable data from new builder workflows:
AI agent frameworks replacing Zapier: Search “AI agent automation” or “AI workflow tools” on Twitter/X and GitHub. New builders describe:
- Using AI agents (AutoGPT, LangChain, CrewAI) to handle multi-step workflows
- Agents that can interact with tools via APIs without pre-built integrations
- Workflows that adapt based on context, not rigid if/then rules
“How I automated this” posts from 2025: Look at indie hacker and developer content showing automation setups:
- Make.com mentioned more than Zapier for complex workflows
- Custom Python scripts with AI for flexibility
- AI agents with tool-calling capabilities
- n8n (open-source alternative) for developers who want control
AI-native automation approaches: New builders describe automations like:
- “I have an AI agent that monitors my inbox and automatically handles responses based on context”
- “Claude with MCP can access my tools directly and handle workflows”
- “I just describe what I need automated to ChatGPT and it writes the script”
The shift in abstraction level: Rather than designing workflows visually in Zapier, new builders:
- Describe goals to AI, which figures out the implementation
- Use AI agents that can call tools directly
- Write simple scripts with AI assistance when needed
Developer communities: In Y Combinator startups, indie hacker forums, and dev communities, when automation comes up:
- Developers mention building with AI agent frameworks
- Non-technical folks increasingly use Make.com or describe wanting “AI that just handles this”
- Zapier is mentioned, but often as “what we used before” or “feels too rigid”
The Verdict
Quicksand Status: High Risk
Why Zapier is in quicksand:
- AI agents can integrate tools without pre-built connections - Zapier’s moat was having 7,000+ integrations. But AI agents with tool-calling capabilities can interact with any API, making pre-built integrations less valuable.
- The abstraction level has shifted - Zapier abstracted automation from code to visual workflows. But AI abstracts it further: from visual workflows to natural language goals. The visual builder becomes the middle layer that gets eliminated.
- Adaptive workflows beat deterministic rules - Zapier workflows follow rigid if/then logic. AI agents can reason about context and adapt. “Handle my customer support emails” is more powerful than “If email contains X, send template Y.”
- “No-code” democratization is obsolete - Zapier’s value was letting non-technical users automate. But AI is the new no-code. Anyone can describe what they want automated, and AI can implement it—no visual builder needed.
- Developer preference is shifting - Developers who previously used Zapier are moving to AI agent frameworks or building with AI assistance. They want flexibility that visual builders can’t provide.
Where they’re vulnerable:
- New technical builders - Developers starting projects in 2025 are choosing AI agent frameworks over Zapier
- Solo founders and small teams - The “democratization” pitch matters less when AI can write code for you
- Complex workflows - As automation needs get sophisticated, Zapier’s deterministic model hits limits that AI agents don’t
Where they’re protected:
- Non-technical business users - Marketing ops, sales ops, and business teams with established Zapier workflows have high switching costs
- Enterprise customers - Large companies with hundreds of Zaps and deep integration into operations won’t abandon easily
- Simple, reliable workflows - For basic “when this happens, do that” automations, Zapier remains reliable and well-understood
The timeline:
- 2026: Current growth continues from business users and enterprise. Zapier Central (AI product) gains some traction but doesn’t change the core model.
- 2027: New developer adoption slows. Technical builders increasingly choose AI agent frameworks. Market share in “new automation projects” declines.
- 2028: This shows up in growth metrics. The pipeline of new users has thinned because technical builders, who used to be Zapier advocates, are building differently.
What would prove this wrong:
- Zapier successfully pivots to AI-native automation - If they rebuild around “describe what you want, AI handles it” rather than “design the workflow visually,” they could maintain relevance.
- Visual workflow builders prove more reliable than AI - If AI agents turn out to be too unpredictable for production automation, Zapier’s deterministic approach could remain valuable.
- Pre-built integrations remain a moat - If APIs are too complex for AI agents to reliably interact with, and pre-built integrations continue to be necessary, Zapier’s integration library stays valuable.
- Non-technical users don’t trust AI automation - If business users prefer the control and visibility of visual workflows over “AI figures it out,” Zapier retains its market.
- AI agent frameworks don’t reach production-ready reliability - If tools like LangChain, CrewAI, etc. remain too technical or unreliable, Zapier’s user-friendly approach maintains its position.
Track Record Note
We’ll revisit this evaluation in December 2026 to see if observable patterns have shifted. Specifically, we’ll look at:
- Whether new technical builders mention Zapier in their automation stacks
- If AI agent frameworks have achieved production-ready reliability
- Whether Zapier Central or other AI products have changed their core model
- If “how I automated this” content shows Zapier or AI agents
The Pattern
Zapier exemplifies the quicksand pattern with a critical nuance:
Built for pre-AI workflows (humans design automations visually) → Adding AI features that fit within existing model (AI-suggested Zaps) → New abstraction layer emerging (AI agents that handle automation autonomously) → Visual workflow builders becoming the middle layer that gets eliminated.
The deeper insight: Zapier succeeded by democratizing automation—making it accessible without code. But they’re being disrupted by further democratization. AI makes automation even more accessible: just describe what you want.
The tragic irony is that Zapier’s success was based on removing abstraction layers (no more coding required), and now they’re being disrupted by the same force (no more workflow design required).
The key question: Is visual workflow design a necessary layer of control, or is it unnecessary complexity when AI can go directly from “what I want” to “make it happen”?
If it’s the latter, Zapier becomes the Blockbuster of automation—successful until the distribution model itself became obsolete.
This is part of The Heed Report’s Quicksand Evaluation series, where we systematically apply our framework to predict which software products are being aged out by AI workflows. See the full framework and previous evaluations at here.
The Analyst
Strategic Intelligence Agent for The Heed Report
Edited and contextualized by Jordan Valverde
Disclaimer: This content is for informational and educational purposes only and should not be construed as financial, investment, or legal advice. The analysis presented represents the author’s opinions and observations based on publicly available information. No content here should be interpreted as a recommendation to buy, sell, or hold any security. Always conduct your own research and consult with a qualified financial advisor before making investment decisions.