Case study · 2026 · AI · Web · Custom software
Describe in plain language what you want automated.Built an AI flow generator for n8n. Two minutes later a working automation drops out, ready to run, no engineer needed.
n8n AI Workflow Generator is a chat-first tool: type the automation in plain language, get back an import-ready n8n JSON with the right nodes, credential slots, and test data. The studio shipped the FastEmbed template index, the Chainlit chat UI, the iterative-refinement loop, and the export pipeline.
- Python
- TypeScript
- Greenlet
- FastEmbed
- aiosqlite
- Chainlit
Inside the build
Overview
- <2 min
- Idea → import
- EN/HU
- Bilingual prompt
- Local
- aiosqlite history
- Vector
- FastEmbed template search
A Chainlit-based chat UI where you describe your workflow ('when a new lead lands in Typeform, send a Slack message and log to Notion'), the AI produces real n8n JSON with the right nodes, credential slots, and test data. Import-ready. Trained on many templates with fastembed vector search to find the closest-matching pattern.
What shipped
What it does
- Natural-language input · Hungarian and English both supported
- Export · ready-to-import n8n JSON
- FastEmbed vector template search · starts from the closest match
- Chainlit chat UI · iterate in a conversation
- Local aiosqlite · conversation and template history stay on your machine
The problem
- n8n is huge · browsing the node catalogue takes hours
- The template gallery is limited · rarely fits your exact case
- Building a custom workflow is just node-parameter wrangling
- ChatGPT-generated JSON rarely imports cleanly
Why it matters
- Workflow idea → running n8n flow in 2 minutes
- Non-technical PM can mock valid n8n JSON
- Example-driven · modernise legacy workflows fast
- Chat-based iteration · refine as you go
How it shipped
- 01 · BRIEF
Why does ChatGPT-generated n8n JSON never import?
Diagnosed the failure mode: ChatGPT lacks node-schema awareness and credential semantics. We built a vector-search-first generator that always starts from a known-good template, then refines.
- 02 · BUILD
Chainlit + FastEmbed + iterative refinement.
Chainlit hosts the chat, the user describes the flow, FastEmbed retrieves the closest matching template, the LLM patches it to the user's spec. Iteration is conversational · 'add a Notion node after the Slack message' just works.
- 03 · SHIP
Local-first deploy · history stays on the user's box.
aiosqlite holds the conversation + template history locally · no external persistence, no leakage of the user's automation logic. Export button drops a JSON file ready for n8n's import.
Stack
Chat
Chainlit chat UI · iterate in conversation
Same UI for first draft and refinement · users don't switch contexts to tweak.
Search
FastEmbed template index
Vector search over a curated template gallery · the generator always starts from a known-good base.
Export
Import-ready n8n JSON
Output is validated against the n8n schema before download · 'JSON copies cleanly into n8n' is a launch gate.
Local
aiosqlite-only persistence
Conversation + template history stays on the user's machine · no SaaS lock-in for sensitive automation logic.
Case study
“Our project managers used to wait days for an engineer to build them an automation. The studio built a tool where the manager just describes what they want in plain language, and two minutes later a working flow drops out, ready to run. Engineers only check the changes. About two engineering hours saved per workflow, and the team is happier.”
