Step 2: Connect Google Sheets

Pull available leads and prepare the batch processing logic.

1. Open Retrieve Lead List and set Document to [YOUR_ID] and Sheet to Filtered Leads (from the list fields).
2. In Retrieve Lead List, keep the filter where Status equals Available.
3. Configure Limit Lead Fetch with Max Items set to 3.
4. Ensure the path flows Current Time Stamp → Retrieve Lead List → Limit Lead Fetch → Iterate Lead Batches.

Credential Required: Connect your Google Sheets credentials in Retrieve Lead List, Update Filtered Leads, Append Sent Leads, and Archive Draft to Sheet.

Step 3: Set Up AI Email Drafting and Parsing

Generate structured email drafts using Gemini and parse them into Subject and Body fields.

1. In Email Draft Generator, set Prompt to Generate a personalized cold email based on the lead data provided. and keep Has Output Parser enabled.
2. Connect Gemini Chat Engine as the language model for Email Draft Generator.
3. Attach Parse Email Structure to Email Draft Generator as the output parser, with JSON Schema Example set to { "Subject": "Subject Line", "Body": "Body Text" } and Auto Fix enabled.
4. Connect Gemini Parser Model as the language model for Parse Email Structure.

Credential Required: Connect your Google Gemini credentials in all Gemini nodes (grouped across 6 nodes, including Gemini Chat Engine, Gemini Parser Model, Gemini Update Model, Gemini Archive Model, Gemini Revision Model, and Gemini Revised Parser).

Step 4: Configure Approval Routing and Parallel Outputs

Send drafts to Telegram for approval while archiving the draft data in parallel.

1. Set Telegram Approval Prompt to Operation sendAndWait and Message to =Subject:{{ $json.output.Subject }}\n\nBody: {{ $json.output.Body }}, and set Chat ID to [YOUR_ID].
2. Keep the route from Telegram Approval Prompt into Approval Route Switch to handle approve/reject logic based on {{ $json.data.approved }}.
3. Note the parallel flow: Email Draft Generator outputs to both Merge Lead + Draft and Telegram Approval Prompt in parallel.
4. In Merge Lead + Draft, pass combined lead and email data forward to Email Archive Agent.

Credential Required: Connect your Telegram credentials in Telegram Approval Prompt, Telegram Revision Query, and Telegram Batch Notice.

Step 5: Configure Email Dispatch and Send Pacing

Send approved emails and pause between sends.

1. In Dispatch Email, set To Email to {{ $('Retrieve Lead List').item.json.Email }}.
2. Set Subject to {{ $('Email Draft Generator').item.json.output.Subject }} and Text to {{ $('Email Draft Generator').item.json.output.Body }}\n\n--\nCompany Team.
3. Set From Email to [YOUR_EMAIL] and keep Email Format as text.
4. Configure Pause After Send with Amount set to 1 to throttle email sends.

Credential Required: Connect your SMTP/Email credentials in Dispatch Email.

Step 6: Set Up Sheet Updates and Archiving

Record sent emails, update lead status, and archive drafts.

1. In Sheet Update Agent, keep the prompt Update sheets: append to Sent Leads and update status to Sent in Filtered Leads..
2. Set Session Memory Buffer with Session Key {{ $('Retrieve Lead List').item.json['Lead \nID'] }} and Context Window Length 100 to track per-lead memory.
3. In Update Filtered Leads, keep Operation as update and mapping values like {{ $fromAI('Status', '', 'string') }} and {{ $fromAI('Lead__ID', '', 'string') }}.
4. In Append Sent Leads, keep Operation as append and map fields to AI values like {{ $fromAI('Name', '', 'string') }} and {{ $fromAI('Sent_Date', '', 'string') }}.
5. Use Email Archive Agent to send combined data to Archive Draft to Sheet with mapped fields {{ $fromAI('Email_Body', '', 'string') }} and {{ $fromAI('Email_Subject', '', 'string') }}.

AI tool sub-nodes like Update Filtered Leads, Append Sent Leads, and Archive Draft to Sheet inherit credentials from their parent agent nodes. Ensure credentials are added to the parent AI nodes using Google Sheets.

Step 7: Configure Revision Loop (Rejected Approvals)

When an email is rejected, collect feedback and regenerate the draft.

1. In Approval Route Switch, keep the second output (Rejected) connected to Telegram Revision Query.
2. Set Telegram Revision Query to Operation sendAndWait and Response Type freeText with message How to change email?.
3. In Regenerate Email Agent, keep the prompt Regenerate email based on user feedback. and ensure Has Output Parser is enabled.
4. Attach Parse Revised Output as the output parser with schema { "Subject": "Subject Line", "Body": "Body Text" }, and connect Gemini Revised Parser as its language model.
5. Confirm that Regenerate Email Agent routes back to Telegram Approval Prompt for another approval cycle.

Step 8: Test and Activate Your Workflow

Validate each branch and then enable the workflow for production.

1. Click Execute Workflow and start from Manual Launch Trigger.
2. Confirm leads are fetched from Retrieve Lead List and limited by Limit Lead Fetch.
3. Verify that Telegram Approval Prompt sends the draft with the subject/body rendered from {{ $json.output.Subject }} and {{ $json.output.Body }}.
4. Approve once to ensure Dispatch Email sends to the lead’s email and that Pause After Send runs.
5. Reject once to ensure Telegram Revision Query and Regenerate Email Agent loop back correctly.
6. When tests pass, toggle the workflow to Active to run it in production.