n8n vs VectorShift vs Letta, which one fits the agent you want to build?
Pick n8n when your agent mostly moves data between apps on a schedule or a trigger. Pick VectorShift when you want to build an AI pipeline or chatbot over your documents without writing code. Pick Letta when the agent has to remember people and past conversations over weeks. The three tools show up in the same searches, and they solve three different problems.
People often compare them because each one can "build an AI agent". In practice the word agent means something different in each. This article compares them on the questions I ask before choosing a platform: what it is built for, where it runs, what the licence lets you do, and who on your team can maintain it.
The short comparison
n8n | VectorShift | Letta | |
|---|---|---|---|
Built for | Workflow automation across apps, with AI steps | No-code AI pipelines, chatbots and search over your data | Agents with long-term memory |
Who builds | Technical operators and developers, visual editor | Non-developers and developers, drag and drop | Developers, with an agent development environment |
Hosting | Cloud or self-hosted | Cloud | Self-hosted open source, or Letta's cloud |
Licence | Sustainable Use License (fair-code) | Commercial | Apache 2.0 |
Best first project | Route form leads to a CRM with an AI summary | A support chatbot over your help docs | A personal assistant that remembers each user |
n8n: workflows first, AI inside them
n8n is a workflow automation platform with a visual editor and a large library of integrations. You build a workflow from nodes (a trigger, an API call, a condition, a model call), and an AI agent node can decide which tools to call inside that workflow.
Its strength is the plumbing. If your agent has to read from a spreadsheet, call an API, post in a chat tool and update a CRM, n8n already connects to most of those. You can self-host it, which keeps data inside your own servers.
The licence is worth reading. n8n uses its own Sustainable Use License, which lets you use and modify it for your own internal business purposes and for personal or non-commercial use, and allows passing it on to others only free of charge for non-commercial purposes. That covers most companies building internal automation.
I wrote a step-by-step example earlier: Exploring n8n with a Reddit automation.
VectorShift: AI pipelines without code
VectorShift is a no-code and low-code platform for building AI applications. You drag blocks onto a canvas (a knowledge base, a model, a prompt, an output), connect them into a pipeline, and publish it as a chatbot, a form or an API.
Its strength is speed for teams without developers. A marketing or support lead can build a chatbot over the company's documents in an afternoon. It runs as a hosted service, and there is a free plan to try it, with limits on pipelines and runs listed on its pricing page.
Choose it when the AI step is the product, and the integrations around it are few.
Letta: agents that remember
Letta comes from the MemGPT research project at UC Berkeley, and it focuses on one hard problem: memory. A Letta agent keeps a core memory in its context, plus archival and recall memory it can search, so it can remember a user's preferences and past conversations across sessions.
Its strength is long-running, personal agents: a tutor, a support agent that knows the customer's history, or a research assistant that builds up knowledge over time. The framework is open source under Apache 2.0, and it is model-agnostic. Developers work with its API and its agent development environment. See the Letta documentation for the current setup.
Choose it when memory is the feature, and your team writes code.
How to choose in practice
- Write one sentence describing the job: "When X happens, the agent does Y and tells Z."
- If Y is mostly moving data between apps, start with n8n.
- If Y is answering questions from your documents and nobody on the team codes, start with VectorShift.
- If Y depends on remembering a person over weeks, start with Letta.
- Build the smallest version in one day and count how many workarounds you needed. The tool with the fewest workarounds is your answer.
You can also combine them. A common pattern is n8n for triggers and integrations, calling an agent built elsewhere through an API.
Questions people ask
Is n8n better than VectorShift?
They are good at different jobs. n8n is stronger at integrations and self-hosting, and VectorShift is faster for building an AI chatbot or pipeline without code. Pick by the job, using the steps above.
Can I use Letta with n8n?
Yes. Letta exposes an API, so an n8n workflow can send a message to a Letta agent and use its reply in the next step. n8n handles the trigger and the apps, and Letta keeps the memory.
Which one is free?
n8n and Letta can be self-hosted at no licence cost, and you pay for hosting and model calls. VectorShift has a free plan with usage limits. All three offer paid cloud plans.
More builders sit in the No-code builder and Workflow Automation categories on AI Agents Live.
