Low-code, no-code, and conversational support bot and AI agent building

On April 3, 2026 we shipped the Macha AI Agent Builder to collapse that, and this post is the proper write-up of what it does. You describe the agent you want in plain English — "triage incoming Zendesk tickets, tag the billing ones, and draft a reply for everything else" — and Macha builds it: the instructions, the tools wired to your actual connected integrations, the triggers, any sub-agents, and the model. You get a real, fully configured agent you can run, inspect, and edit — not a stub you have to finish by hand. It's live now, and it's available on all plans.

Hey Z-Admin, Zendesk Relate is over, and if you missed it, here’s a quick summary of the most relevant announcement 👇

The main topic was something Zendesk is calling the Autonomous Service Workforce. What does that mean? In simple terms, Zendesk is moving from traditional bots that only deflect tickets to AI agents that can work across channels, understand context, and take action.

✅ Agent Builder: A no-code interface to create, test, and optimize your own AI agents, adapted to your policies, workflows, data, and business logic.

✅ Omnichannel AI Agents: AI agents will be able to operate across messaging, email, voice, ChatGPT, Gemini, and other environments while keeping context across channels.

✅ Copilots for different roles: Not only for agents. Zendesk is also expanding Admin Copilot, Knowledge Copilot, and Analyst Copilot to help detect issues, improve workflows, find knowledge gaps, and analyze trends.

✅ Action Flows + MCP + connectors: Zendesk wants AI to do more than answer questions. The goal is for AI agents to execute actions across your external systems.

✅ Resolution-based pricing: Zendesk is moving toward a model where customers pay for verified resolutions, not just usage or volume.

What do you think?

I built a support bot last week without opening n8n once.

Described what I needed out loud to Claude Code. It pulled the n8n docs, checked the Asana task where the client described requirements, built the workflow through n8n-mcp.com, and deployed it.

Then I said: test it.

It tested. I left one node disabled so responses wouldn't go live, and let executions accumulate for two days. Then I said: review the executions.

It analyzed hundreds of runs and came back with: "We missed several queries. Someone asked 'where is my invoice' instead of 'where is my order' and we didn't handle that."

I said: update the workflow.

It updated it.

Two years ago I was spending 95% of my time dragging blocks in the n8n UI. Today the ratio is almost inverted. I work in Claude Code, talk to my computer, and barely touch the canvas.

n8n didn't become less important. It stopped being the place where I build automation and became the place where I run it. The building moved to the conversation layer.

Hundreds of workflows built this way so far. Most of them voice-to-deploy.

Curious where other builders draw the line — what's the part of workflow building you still want to do by hand?

I run a small business and recently explored AI for customer support while trying to answer a simple question: what is the best AI for customer support? I tested a few options and wanted to share my experience in case it helps someone else looking into it.

My goal was straightforward: reduce repetitive support work, respond faster, and avoid making the customer experience feel robotic or broken. I didn’t need anything enterprise-heavy - just something I could actually set up and maintain without a full tech team.

I tested a few tools: n8n, Moveworks, and nexos.ai.

n8n was interesting because it’s flexible. You can build almost anything if you’re willing to wire it together yourself. The downside (for me) was exactly that - it felt like building everything from scratch. It’s great if you enjoy automation and logic flows, but less ideal if you want something ready to use quickly. Also one pretty big decisive factor is its price, it is quite big for a small business.

Moveworks felt polished, but also very enterprise-focused. It seemed powerful, but a bit too much for a small business like mine. I also felt like I didn’t have as much control over how conversations actually flow.

I ended up choosing nexos.ai for now. What I liked is that it sits somewhere in the middle - not too DIY, but not locked down either. I could set up flows, automate responses, and still tweak things without needing to be technical. It also felt smoother in real customer conversations, which matters a lot when evaluating the best AI agents for customer support.

If you’re trying to figure out what is the best AI for customer support, I’d say the most important factor is how easy it is to set up and maintain but only if it still has the features you actually need. I came across a comparison table of AI tools that helped me narrow things down and find the ones I’ve already tested. At the end of the day, most tools can automate replies; the real difference is how much time you spend managing the system vs. letting it run.

Would be great to hear what others are using - especially other small business owners. Did you go full custom, or find something more plug-and-play that actually works long term?