Hands-On Evaluation of n8n and Peer Products for AI Automations

Harshal
Oct 30
6 min read

A Product Immersion Deep-Dive Example for PM Interviews

Here, I share a deep dive into a product memo example I created for n8n a few months ago, when I interviewed with them. Fast-forward, I got a job offer and will join them in Q4 2025. Since this doesn’t have n8n employee inputs, the information is not accurate, but I hope it gives you ideas on the approach.

In earlier posts, I shared a framework to write a case study or product memo when you interview for Product Management roles with companies.

I did an exploratory evaluation of n8n and peer products, focusing on their automation and AI capabilities.

These insights may now be outdated, as I created them before joining n8n. My goal here is to show you a deep-dive example for your Product Management interviews.

Scope Of Hands-On Evaluation

I did a hands-on evaluation of n8n and peer products, focusing on their automation and AI integration capabilities. My testing covered:

n8n: Primary focus. Tested a template to scrape company news and built a RAG chatbot using a Pinecone vector store. Tested triggers, API integrations, OpenAI and Gemini AI agents, error handling, and conditional logic.
Zapier: Tested as a benchmark comparison as a leading no-code integration platform. I looked at examples of copying meeting invites from one calendar to another.
Node-RED: Evaluated its capabilities for smart home.
Tines: It did not offer self-service sign-up. I paused after watching its intro video.
LindyAI: Tested its AI builder by creating a workflow to scrape news of a company and email me. Used pipedream to handle credentials.
Google Opal: Opal was a recent launch by a tech giant.
AgentAI: Hubspot’s co-founder supported this project to create autonomous agents.

n8n: The Good

Template-based approach: Ready-made templates like "Scrape Recent News About a Company" provide quick starting points for automation.
Integration capabilities: Seamless connection to Google accounts and other services, similar to Zapier experience.
AI integration: Ability to add AI summarization capabilities for transforming lists of news into cohesive summaries.
Flexibility in triggers: Multiple ways to trigger workflows (calendar events, manual text input).
Advanced capabilities: Support for complex workflows like RAG chatbots using Google Drive, Gemini, and Pinecone.
Visual workflow builder: Allows seeing the entire automation flow and connections between nodes.

n8n: The Bad

Template customization challenges: Adapting templates to specific use cases (like changing calendar filtering) requires significant modification.
Error diagnosis complexity: Troubleshooting issues like the "get news" node not working requires detective work to identify problems.
Variable handling: Some issues with variable evaluation and handling empty responses require additional logic nodes.
API setup requirements: Need to obtain external API keys (e.g., from newsapi.org) to make certain templates work.
Documentation gaps: Sometimes steps are mentioned in documentation but easy to miss, like adding email as a "test user".
Complex authentication: Multi-step process to enable APIs, set up OAuth, and configure credentials.

n8n: The Challenges

High time-to-value: "A lot of pain to reach first value" even when using templates sometimes, for example, with the RAG chatbot setup.
Dependency on multiple services: Your automations are between services. Complex workflows require juggling multiple external services and their configurations.
Node interdependencies: Missing or misconfigured nodes (like the "setup" node) can break entire workflows without clear error messages.
Debugging limitations: Challenging to identify which component needs modification when something doesn't work.
Redirection loops: It was frustrating to be repeatedly redirected to templates even after installation. It also seemed that the templates page was unaware of my user ID as the templates shown were not customized to my situation.
Learning curve: Understanding how to properly configure and connect nodes requires significant effort and troubleshooting.

Hands-on with n8n

Below, I'll share some screenshots of my experience. I tested n8n using their template library and found the credential setup surprisingly straightforward - all credentials could be added from a single page. This was helpful. Yet, we are trying to simplify a complex item: API integrations. They still typically require complex token creation steps, which can be challenging.

the connection to google accounts was seamless.

The AI assistant helps me learn confusing components, and I appreciate seeing JSON expression evaluations. This product appears designed for technical users based on its information-dense interface. The various panes labeled "Input," "Output," "Settings," and "Parameters" use terminology more common in technical than business environments.

i followed the AI instructions and it worked fine.

ended up creating a completely parallel path to the template

For a widely used product like n8n, 4,000-5,000 templates feels insufficient. The platform needs 10 times more templates, which n8n could generate using AI assistants. For example, if you have one template which uses the Gmail trigger, another template can use the email trigger. If you have one template which uses the OpenAI chat model, another can use the Gemini chat model. These kinds of variations are important because n8n is catering to the long tail. It will also help in discoverability through SEO.

ready and running with AI as I did notwant a table of links but some explanation of it

Model selection remains challenging! You can simplify building AI but some of core AI decisions remain hard.

Not clear which model to choose or change settings in the pinecode vector store

When exploring the n8n template library, I searched for a common operational automation: onboarding new company employees. Despite this being a frequent use case mentioned in tutorials and blogs, the search didn't yield comprehensive results. I found templates with partial overlap but nothing that fully addressed this workflow. This experience highlighted why the template library needs at least 10× more templates to effectively serve diverse user needs and needs a natural language search.

template search was not generative or in-depth in n8n template library.

Node-Red in Home Assistant

I use Node-Red in Home Assistant.

A huge canvas for multiple workflows in node-red

I wrote more about my smart home here, and will write more on it later.

Larger smart home workflows for kids sleep and wakeup

Zapier

I've used Zapier extensively because it consistently appears in search results for specific automation needs. Recently, I deactivated several automations after reaching the free plan limit. However, I can still review its capabilities for comparison.

LindyAI

I tried a prompt on LindyAI:

i want to build an agent that receives an email on gmail with name of company in title and scrapes news for that company in last 2 weeks and summarizes and links to the sources in an email.

It repeatedly asked clarifying questions before building anything.

gave me a list of credentials I need to set up.

LindyAI uses Pipedream to handle login, which is interesting because Pipedream is also a workflow automation platform.

similar approach to nodes as n8n’s “AI agent” node.

It guided me through testing each step.

But the execution seemed very slow. I liked that it could branch based on existence of news or not. I deployed it but there was some error, which I could not debug.

Reflections

Over the past year, I've tested competitive products to understand my users better. This practice helps me consider multiple solutions to user needs. During interviews with companies, this approach gave me valuable industry insights.

Testing competitor products helps product managers understand users and identify effective UI patterns.

Rest Of The Product Memo

You can read rest of the product memo here.