
N8n rolls out AI Assistant preview for Business/self-hosted, cuts Enterprise
n8n's AI Assistant status changed across plans: on the Business plan and for self-hosted deployments, it moved from "coming soon" to "(preview)" with BYOK availability.
N8n’s pricing page · Change Detected
WHY IT MATTERS
The open question is whether Enterprise access returns once usage and cost patterns settle, and whether BYOK becomes n8n's default packaging for future AI features rather than a transitional step. Watch whether n8n reinstates an Enterprise AI Assistant commitment in the next few release cycles, and whether other automation platforms adopt BYOK gating to limit their own AI cost exposure.
BY THE NUMBERS
- Features typically cascade upward, with top tiers first
- Here the opposite happened: preview access advanced for Business and self-hosted while Enterprise's promised access was withdrawn outright
THEIR RATIONALE
n8n frames this as advancing AI Assistant into hands-on preview for Business and self-hosted users via a bring-your-own-key model, letting those customers test the feature without n8n absorbing model-inference costs. Pulling the prior "coming soon" promise from Enterprise suggests n8n is re-scoping who the feature is ultimately for, at least for now.
THE SIGNAL
Features typically cascade upward, with top tiers first. Here the opposite happened: preview access advanced for Business and self-hosted while Enterprise's promised access was withdrawn outright. Paired with the BYOK requirement, this reads as n8n testing AI Assistant demand and usage patterns before committing to who bears inference cost and at what tier, a hedge increasingly common among workflow-automation vendors layering AI into existing plans.
WHAT PEOPLE ARE SAYING
Just got n8n’s new AI Assistant running on my self-hosted instance 🔧 No more manually tracing broken workflows, just describe the issue and it debugs with you, running code in its own sandbox. Docker + Traefik + a lot of patience, but worth it. #n8n #Automation #SelfHosted #AI
View originaln8n just killed the complexity of self-hosted AI. No more 6-step manual configurations. No more technical setup headaches. Here’s the n8n AI play 👇 → One-command deployment for servers → AI Assistant builds workflows from text → Full control over your private data → Swap any AI model (OpenAI/Anthropic) → Automate leads and admin on autopilot Stop renting the cloud and start owning your tech. The future of automation is local and AI-powered. Want the full guide? DM me.
View originalNo shit, the new AI assistant in n8n is a game changer. One catch if you want to use true self-hosted open-weights models (like Qwen 2.5 on vLLM or Modal) instead of paid OpenAI keys: n8n’s Vercel AI SDK hits /v1/responses and throws a schema mismatch error. Built an open-source FastAPI translation proxy to fix the endpoints and live SSE streaming:
View original