Skip to content
ExpertEnterpriseChatbots & Conversational AIWebhook trigger32 nodes

Automate an assistant workflow: Notion for voice

Fires the moment a webhook arrives across Notion, Telegram and Slack, hands the decision to an AI agent with its own tools, then branches on the check before anything is written. Built for voice, around a "Keep Valid Conversations" step, in 32 documented nodes.

Automate an assistant workflow with Telegram without manual work for Voice. An original, ready-to-import n8n workflow built around the 'AI Agent' pattern for customer support. It fires on a webhook and orchestrates 32 nodes across Notion, Telegram, HTTP / REST API. 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. The result is a hands-off process that just works.

Key benefits

  • Keeps Notion and Telegram in step for voice, without anyone re-typing a row
  • Safe to re-run: records already in Notion or Telegram are skipped, across all 32 nodes of this voice build
  • A failed Notion or Telegram call is retried rather than dropped, at any of the 32 nodes in this voice build
  • Branches rejoin before Notion, so a voice run reports once rather than 32 times
  • Every voice run leaves an audit row behind, alongside Notion

Use cases

  • Teams keeping Notion and Telegram aligned by hand for voice
  • Voice work that needs a judgement call per item before it reaches Notion
  • Pushing voice events into Notion and Telegram the moment they land, not on the next poll
  • Voice work where "Keep Valid Conversations" is still a manual step before Notion

Integrations

NotionTelegramHTTP / REST APISlackGoogle SheetsOpenRouter

Workflow preview (32 nodes)

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

On Telegram EventTelegram TriggerKeep Valid ConversationsFilterReason & Act (AI Agent)AgentOpenRouter Chat ModelLm Chat Open RouterLookup API DataTool Http RequestThink It ThroughTool ThinkCustom Logic ToolTool CodeRun CalculationsTool CalculatorConversation MemoryMemory Buffer WindowShape Agent OutputSetConfident Enough?IfDeliver Answer + Log to…Google SheetsEscalate to Human QueueTelegramJoin OutcomesMergeLoad Previous SnapshotNotionCompute Drift vs Snapsh…CodeEnrich with Reference D…Http RequestFold In Enrichment Fiel…SetStamp Completion TimeDate TimePersist Run MetadataExecution DataSerialize Run Output (C…Convert To FileArchive Row SnapshotNotionOffload to Post-Process…Execute WorkflowRecord Sub-workflow Res…Execution DataFingerprint for Idempot…CodeDrop Already-Processed …Remove DuplicatesSummarize Run StatsSetAppend Audit Row (Notio…NotionSend Summary to SlackSlackCapture Failure ContextSetEscalate Failure to Tel…TelegramPark in Dead-Letter Que…Notion

What's inside (32 nodes)

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

AgentCodeConvert To FileDate TimeExecute WorkflowExecution DataFilterGoogle SheetsHttp RequestIfLm Chat Open RouterMemory Buffer WindowMergeNotionRemove DuplicatesSetSlackTelegramTelegram TriggerTool CalculatorTool CodeTool Http RequestTool Think

Agent tools included (4)

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.

  • Custom Logic Tool · Custom code
  • Lookup API Data · HTTP request
  • Run Calculations · Calculator
  • Think It Through · Scratchpad reasoning

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

Architect level

The deepest tier - sub-workflow delegation, state snapshots, and full observability.

  • Wiring an AI Agent to a chat model, memory, and multiple tools
  • Wiring an AI Agent to a chat model and a set of tools
  • Writing a system message that controls when the agent reaches for each tool

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 (Notion, Telegram, HTTP / REST API, Slack, …). 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.