I love the flexibility of the self-hosted version. It allowed our team to automate sensitive customer data workflows without sending that data to a third-party cloud, which was a dealbreaker for us on other platforms. Review collected by and hosted on G2.com.

What I find most challenging is the steep learning curve for non-developers. While the power is there, the UI can feel overwhelming when trying to map complex JSON data without a background in Javascript. Review collected by and hosted on G2.com.

This is one of the best and easiest ways to implement AI workflows without needing extensive technical knowledge or a deep understanding of AI concepts. Interestingly, it also helped me learn and apply AI concepts such as Agents and LLMs. Review collected by and hosted on G2.com.

The licensing for this application can be somewhat confusing and tends to be on the expensive side. While the software is still evolving, it has yet to fully mature for large-scale enterprise deployments, particularly in areas where governance, privacy, and security are critical—especially given the current landscape of AI. Review collected by and hosted on G2.com.

The power it gave me to build AI workflows when other tools had limitations. Self hosting the community edition is a big + for my solo projects.

There is a big of a learning curve but it is actually healthy as it helped me learn concepts like JSON, javascript, to really understand how integrations work Review collected by and hosted on G2.com.

So there are a couple of things that could be improved. For the AI integrations, we still aren't able to fully customize the settings of the AI models that we use. For example, GPT-5 has been released quite a while ago with new parameters like minimal reasoning or verbose, which are elements we still cannot edit when using OpenAI as a chat model. I must say that the cloud instances have very limited power and crash a lot when processing lots of data. Review collected by and hosted on G2.com.

What I like most about n8n is that it gives me the power of a low-code tool without boxing me in. The visual workflow builder is easy to pick up, so getting my first automations live didn’t take long. At the same time, it still lets me build fairly complex logic when I need it. Once a workflow is set up, it just runs in the background and stays out of the way, which means it’s something I actually use regularly instead of only testing once and then forgetting about it.

I also appreciate how many features and integrations are available out of the box. Connecting different tools is usually straightforward, and when something isn’t supported directly, calling custom APIs is still simple enough to do. The documentation and community support have been helpful as well, especially when I’m trying to tackle something a bit more advanced.

Overall, n8n hits a sweet spot between ease of use, flexibility, and straightforward implementation and integration—something I haven’t really found in other automation platforms. Review collected by and hosted on G2.com.

I wouldn’t say I “dislike” n8n, but it does have a few rough edges. The learning curve can feel pretty steep at the beginning, especially if you’re not already comfortable with APIs or other automation tools. Some nodes and error messages come across as quite technical, so getting more complex workflows to run exactly the way you want can take some trial and error.

Overall, performance and stability are solid, but I do occasionally run into small bugs or editor quirks that break my momentum. The documentation has improved a lot, which I appreciate, yet there are still times when I wish it included more real-world examples or clearer guidance for advanced use cases.

None of this is a deal-breaker for me, but these are the main areas where I think n8n still has room to grow. Review collected by and hosted on G2.com.

I love the UI, best tool on the market, in comparison to Make or Zapier. We can self-host it, setting the limits far higher.

The number of integration is far enough, and to play with the HTTP request additionnaly is nice enough.

The fact that we can easely integrate javascript is very helpful.

Over the recent years, n8n has helped us getting deeper into using IA, which is absolutely necessary. Review collected by and hosted on G2.com.

The online documentation is not good enough, fortunately n8n gave us an AI RAG tool to discuss with the documentation, making it way better.

Also, it feels like we miss templates, to understand how people used specific tools. For now it feels like the template are pretty much the same, too simple for real company usecases Review collected by and hosted on G2.com.

Oh, ease of use. Unlimited posibilities. Having the code nodes gives incredible flexibility to do whatever I need. What it used to take FOREVER to do with MULE like 20 years ago, now it takes you minutes to accomplish. Having so many nodes allowing you to integrate with plenty of systems is super useful. Being able to plug AI into the flow to make decisions is also super valuable. Review collected by and hosted on G2.com.

I wouldn't say I dislike something about N8N, but I have ran into issues when setting up nodes, as the documentation in Chat GPT is not accurate.

But that is not N8N's fault. Review collected by and hosted on G2.com.

There are a whole lot of nodes to work with for every service and workflow. It’s open source, which makes it possible to self-host on our servers and connect it to a custom domain. Automation can be implemented to cover more than enough business requirements. Review collected by and hosted on G2.com.

Some people may find it difficult to authenticate with the service providers. Notion, Google, and similar service nodes can also be challenging for a beginner to set up. Review collected by and hosted on G2.com.

I really like n8n for several reasons. It’s incredibly flexible and lets you build complex, multi-step workflows that fit your exact needs, which is great if you want to connect different APIs and customize logic. I also appreciate that it’s open-source, so you can self-host, extend it with custom nodes, and have full control over your workflows. It’s very cost-effective, especially for smaller teams or personal projects, and the community is active, so there’s a lot of shared knowledge, examples, and support. On top of that, it balances no-code and low-code features nicely, allowing people to build useful automations without needing to write everything from scratch, while still giving advanced users the option to code where needed. Review collected by and hosted on G2.com.

Some popular apps don’t have built-in nodes, so integrations can require extra setup. Performance can lag with very complex or data-heavy workflows, particularly if the instance isn’t optimized, but the flexibility, control, and cost-effectiveness tend to outweigh these challenges for someone focused on AI automation and custom solutions. Review collected by and hosted on G2.com.

i like about n8n is i can use deterministic workflows for reliability and AI agents only where reasoning is needed. That’s probably more practical than the “fully autonomous agent” hype everywhere right now Review collected by and hosted on G2.com.

One thing n8n can definitely improve is the deployment experience. Building workflows is easy and powerful, but deploying and maintaining n8n in production is still quite complicated.

Setting up Docker, SSL, reverse proxy, Redis, workers, environment variables, and scaling takes a lot of technical effort, especially for small teams or non-DevOps users. Sometimes even small configuration mistakes can break executions or webhooks.

I’d love to see:

simpler one-click deployment

easier cloud/VPS setup

built-in scaling and monitoring

simpler backup & restore options

better error debugging/log visibility

smoother setup for queue/workers mode Review collected by and hosted on G2.com.

I like the out-of-the-box features of n8n, especially its integration capabilities with the majority of popular software available online. The test executions and the history of executions make debugging really easy. Also, working with the MCP offered as a third-party solution is very straightforward for creating new workflows, particularly with the usage of large language models. Review collected by and hosted on G2.com.

Well, until we set up the automated error handling, notifications to Teams, there were occasions where we missed some workflow was working, but then, for example, some API format have changed or new documents have arrived and was not properly handled by our workflows. And then the workflow was dead for a couple of days. So maybe some things about error handling that could be built in, like email notifications. And then the backup solution, we had an opportunity where the virtual machine that hosts n8n was deleted, and then we struggled a little bit with reverting the n8n configuration from the backup. Review collected by and hosted on G2.com.