On Monday October 6, 2025, OpenAI announced AgentKit and Agent Builder at DevDay, and it’s genuinely exciting. The ability to visually design multi-agent workflows, connect to enterprise data sources, and deploy production-ready agents represents a massive leap forward for developers building AI systems. Sam Altman called it “Canva for agents,” and the demos show just how powerful this new toolkit can be.

But here’s the thing: developers represent less than 1% of the U.S. workforce, while knowledge workers make up 42%.

The Democratization Paradox

Think back to 2022. Before ChatGPT, AI was primarily the domain of ML engineers and data scientists. Then ChatGPT arrived and changed everything. Suddenly, marketers could draft campaigns, sales teams could research accounts, and operations managers could automate reporting, all through natural language with no coding required.

GenAI democratized access to AI capabilities for millions of knowledge workers. It removed the technical barrier between “I have a problem” and “AI can help me solve it.”

Agent Builder, for all its power, doesn’t fully complete the democratization of agentic workflows.

Looking at the announcements from today, it’s clear that Agent Builder is designed for developers, solution architects, and technical teams working with APIs. You need to understand:

• How to design workflow logic with nodes and connections

• What do reasoning effort and output formats mean

• How to configure evals and guardrails

• When to use different models and parameters

These are valuable capabilities for the roughly 1.9 million software developers in the U.S. However, they’re not accessible to the 71.5 million knowledge workers, including compliance managers monitoring policy changes, customer success leaders preparing quarterly reviews, and operations teams coordinating cross-functional workflows.

What About Other No-Code Agent Builders?

Yes, there are dozens of no-code agent builders available today. Tools like Voiceflow, Botpress, n8n, and Microsoft Copilot Studio have made building AI agents more accessible than ever before.

But there’s a critical distinction most of these tools miss: they’re still workflow builders dressed up as no-code platforms.

Most no-code agent builders give you a visual canvas instead of a code editor. You drag nodes, connect them with arrows, configure triggers, set conditional logic, and wire up integrations. It’s simpler than writing code, yes. But you’re still architecting the solution, still thinking in terms of “if this, then that,” still mapping out every step of the process.

That’s perfect for operations teams, technical product managers, and business analysts who think in terms of systems and workflows. But it’s not accessible to the subject matter experts who just want to describe the outcome they need.

A Different Approach: Natural Language Agent Building

At Larridin, we believe there’s a better path. Instead of asking business users to design workflows, let them describe what they need done, the same way they’d explain it to a colleague.

Here’s what we mean:

• Traditional no-code approach: Open visual builder → Add trigger node → Configure data source connector → Set up comparison logic → Define conditional branches → Add approval step → Connect to output system → Test workflow → Deploy
  
  

• Natural Language (conversational) approach: “Create an agent called ‘Daily Morning Briefing.’ This agent should work daily as an analyst who searches the company’s knowledge base and sends a summary of key insights via email.”

The conversational approach expects the system to handle the orchestration, data connections, and workflow logic. The user focuses on the outcome.

Why This Matters for Enterprise

The people who best understand your business workflows aren’t typically the ones who can (or want to) design state machines. Your compliance manager knows exactly when policies need review and what steps should happen next. Your customer success director knows which signals indicate account health and what actions to take. Your sales ops lead knows how to qualify leads and route them appropriately.

But asking them to translate that knowledge into a visual workflow builder creates the same bottleneck we’ve always had. Either they wait weeks for engineering support, or they become “citizen developers” (which sounds empowering but really means “people doing work outside their expertise”).

We think there’s a better way: agents that work the way business users think.

Building for the Other 99%

Larridin Nexus is available today as a secure AI gateway that connects your teams to approved LLMs with enterprise governance. We provide curated prompt libraries, role-specific templates, and secure connections to your business systems like Slack, Salesforce, Jira, and Google Workspace.

And we’re building agent capabilities designed specifically for business users:

Outcome-focused conversations, not workflow design: Describe what you want accomplished. The system determines how to orchestrate it.

Enterprise context by default: Agents understand your company’s data, systems, and policies without requiring manual configuration of every connector and permission.

Built-in governance: Security, compliance, and data handling aren’t configuration options you might miss. They’re embedded in how the system operates.

Role-specific starting points: Pre-built agent templates for common enterprise needs that you can customize through conversation, not configuration.

Both Approaches Matter

To be clear: we need both paradigms. Developers and technical teams building sophisticated, custom agent systems with tools like OpenAI’s Agent Builder will unlock incredible capabilities. We’re genuinely excited about what teams will build with it. And for ops professionals who think in workflows, visual builders like n8n and Zapier are powerful and appropriate.

But we also need tools that meet business users where they are: describing outcomes, not designing systems.

The AI revolution shouldn’t stop at content generation. Automation, orchestration, and intelligent workflows should be just as accessible to everyone who can describe what they need done.

That’s the future we are building toward.

Larridin Nexus is available today. If you’re interested in learning more or joining our early access program, reach out to our team.