Skip to content
ExpertPremiumDocument ProcessingWebhook trigger23 nodes

Distill documents: PostgreSQL for invoices

Fires the moment a webhook arrives across PostgreSQL, Slack and Google Sheets, hands the decision to an AI agent with its own tools, then branches on the check before anything is written. Built for invoices, around a "Reason & Act (AI Agent)" step, in 23 documented nodes.

Distill documents with PostgreSQL and Telegram end to end. An original, ready-to-import n8n workflow built around the 'AI Summarize & Route' pattern for legal & compliance. It fires on a webhook and orchestrates 23 nodes across PostgreSQL, Slack, Google Sheets. It uses n8n's LangChain AI nodes (Agent/Chain + chat model). Retry-on-fail is enabled on external calls. Import the JSON, add your credentials, and adapt the parameters. Your team sees the right data at the right time.

Key benefits

  • Keeps PostgreSQL and Slack in step for invoices, without anyone re-typing a row
  • A failed PostgreSQL or Slack call is retried rather than dropped, at any of the 23 nodes in this invoices build
  • Bad input stops the run for invoices instead of landing in PostgreSQL
  • Branches rejoin before PostgreSQL, so a invoices run reports once rather than 23 times
  • The invoices AI step is isolated from PostgreSQL, so models swap without touching the other 23 nodes

Use cases

  • Teams keeping PostgreSQL and Slack aligned by hand for invoices
  • Invoices work that needs a judgement call per item before it reaches PostgreSQL
  • Pushing invoices events into PostgreSQL and Slack the moment they land, not on the next poll
  • Invoices work where "Reason & Act (AI Agent)" is still a manual step before PostgreSQL

Integrations

PostgreSQLSlackGoogle SheetsTelegramOpenRouter

Workflow preview (23 nodes)

The template's actual node layout and connections. Node parameters and credential slots unlock with the download.

On Telegram EventTelegram TriggerRequired Fields Present?IfAbort — Malformed InputStop And ErrorReason & Act (AI Agent)AgentOpenRouter Chat ModelLm Chat Open RouterRun CalculationsTool CalculatorSearch Reference WikiTool WikipediaConversation MemoryMemory Buffer WindowShape Agent OutputSetConfident Enough?IfDeliver Answer + Log to…PostgresEscalate to Human QueueTelegramJoin OutcomesMergeSerialize Run Output (C…Convert To FileArchive Row SnapshotPostgresLoad Previous SnapshotGoogle SheetsCompute Drift vs Snapsh…CodeSummarize Run StatsSetAppend Audit Row (Postg…PostgresSend Summary to SlackSlackCapture Failure ContextSetEscalate Failure to Sla…SlackPark in Dead-Letter Que…Google Sheets

What's inside (23 nodes)

A look at the node types this workflow uses. No purchase required — full parameters and credentials are yours after you buy.

AgentCodeConvert To FileGoogle SheetsIfLm Chat Open RouterMemory Buffer WindowMergePostgresSetSlackStop And ErrorTelegramTelegram TriggerTool CalculatorTool Wikipedia

Agent tools included (2)

An AI Agent is only as capable as the tools wired into it. These are read from the template's own graph — every one is a real node in the file you download.

  • Run Calculations · Calculator
  • Search Reference Wiki · Wikipedia

Conversation memory is attached, so the agent carries earlier turns of a session.

From download to running in 3 steps

  1. 1
    Import. In n8n, open Workflows → menu → “Import from File” and pick the downloaded JSON.
  2. 2
    Connect. Add your own credentials on the app nodes - n8n highlights exactly which ones need them.
  3. 3
    Activate. Run it once to test, then toggle Active. The automation is live.

What you'll learn

Production level

Production-grade: idempotency, dead-letter handling, and reconciliation included.

  • Routing classified output into separate action branches
  • Rejoining parallel branches with Merge so the run reports once
  • Moving data between Telegram, PostgreSQL and Google Sheets in a single run

Every node carries its own documentation on the canvas, plus notes explaining why the architecture was built this way, a credential setup guide, troubleshooting, and three practice exercises. Sample data comes pinned to the trigger, so you can hit Execute and watch data flow before connecting a single account.

Common questions

What exactly do I get?

The complete, ready-to-import n8n workflow as a JSON file, delivered instantly after payment. Import it into your own n8n (cloud or self-hosted), add your credentials, and it runs.

How do I import it into n8n?

In n8n, open Workflows, click the three-dot menu, choose “Import from File”, and select the downloaded JSON. Then connect your own app credentials on the highlighted nodes.

Do I need anything else for it to work?

You need your own n8n instance and accounts/credentials for the apps this workflow connects to (PostgreSQL, Slack, Google Sheets, Telegram). No coding is required.

What if it doesn't work for me?

If the file is faulty, won't import, or isn't as described, contact us within 7 days and we'll fix it or refund you - see our refund policy.

Can I modify or resell it?

You can freely adapt and use it in your own or your clients' projects. Reselling or redistributing the template file itself is not permitted.

Original n8n workflow template created and owned by WorkflowCrate. After purchase you receive the ready-to-import JSON file and a license to use and adapt it in your own projects.