Digital
workshopDiscuss a task RU

OWN ENGINEERING SYSTEM / N8N · API · POSTGRESQL · AI

Data monitoring through APIs.

Built collection of financial data from APIs, processing in n8n and storage in Supabase. A Telegram Mini App is used for viewing; AI workflows analyze events.

My role
Data architecture, n8n workflows, APIs and the AI layer
What is inside
Sources, normalization, Mini App and alerts
Result
Data is mapped to one structure and available for monitoring.

Task and my work.

Monitoring data came from different kinds of sources: exchange APIs, wallets and news feeds. The goal was to combine them for my own monitoring and flag events worth a human look.

  • Built the collectors and data processing in n8n.
  • Separated raw responses and diagnostics from the normalized layer the interface reads.
  • Connected Supabase/PostgreSQL to a secured Telegram Mini App.
  • Added AI event analysis with structured output, validation and rate limiting of repeated signals.

Key decisions.

No final state is saved without required data.

Required and optional sources have different rules. Without required data the system stops saving. If an optional source fails, it records a warning.

The AI analyzer runs without server tools.

Command execution tools are disabled for the Hermes analyzer. Its response is validated before any alert; a rate limit holds back repeated signals.

How it works.

  1. Collect data

    API collectors store responses and error details.

  2. Normalize

    A separate workflow builds one structure for the interface to read.

  3. Analyze events

    The AI returns a structured decision; the result is validated before an alert goes out.

Implementation and example.

Screens with fictional data: source status, AI event review and a bot alert. Real balances and access are not shown.

01 / Source status: required and optional sources, warnings and gaps.
02 / AI review: decision, confidence, schema check and the repeat-signal limit.
03 / The alert in the bot and its acknowledgement.

DIAGRAM OF THE IMPLEMENTED SYSTEM

  1. API sources → n8n

    Data collection, raw responses and diagnostics.

  2. Normalizer → Supabase

    Shared data model and a read layer for the interface.

  3. Mini App + AI monitor

    Status view and event analysis with alert control.

Architecture diagram of the project. Live data and connections are not shown.

Project result.

Collection, normalization, the user interface and AI event analysis are in place. Some sources still have gaps in price data. The system is for monitoring; decisions and any actions with assets stay with a person.

Technologies and implementation details

n8n, JavaScript, APIs, Supabase/PostgreSQL, Hermes and a Telegram Mini App. Per the current documentation, the AI layer runs on Hermes. This is not automated trading or a return forecast.

Let's discuss your task.

Tell me the sources, update frequency, processing rules and who should get the result. We will design data collection, error handling and the actions the system should take.

Discuss n8n and integrations ↗︎
PROJECT SCREEN