Application/Build
Built the local foundation for my own AI agent
I moved beyond standalone n8n automations and built the first local foundation for an AI agent I can control, extend and audit.
- Netanel Lacroix
I started with n8n automations and built a local stack around them. That worked well for repetitive tasks, but I wanted more control over how the system thinks, remembers and acts. So I began building my own local AI agent.
The first step was the foundation: a local API as the central entry point, local Postgres for memory and runtime data, n8n as the workflow execution layer, and local or OpenAI models selected according to the task. Approvals and audit logs keep sensitive actions controlled and traceable.
With this foundation in place, I can send commands through Telegram, search and store information, trigger workflows, confirm sensitive actions, and keep a record of what the agent executed. The larger goal is to own the main parts of the system locally: memory, models, tools, permissions, interfaces and the rules around autonomy.
Next, I am giving the agent a stronger brain so it can understand requests, build plans and choose the right tool by itself. LinkedIn update: https://www.linkedin.com/posts/netanel-lacroix_i-started-by-experimenting-with-%F0%9D%97%BB%F0%9D%9F%B4%F0%9D%97%BB-%F0%9D%97%AE-activity-7487409129465233408-X3o7
- local AI agent
- n8n
- Postgres
- AI automation
- governed actions