Skip to content
Stackly by Temaco

Personal project

A message written for the recipient, not a mass send.

The bot takes what is known about a recipient, builds copy that fits their situation and sends it at a scheduled time. The difference from a mass send is that each message genuinely refers to that particular person.

The problem it solves

Blasting one template at everyone works less and less, because people recognise it in two seconds. Writing each message by hand does not scale past a dozen or so a day.

Built with

  • n8n
  • OpenAI
  • Resend
  • Google Sheets

How it works

  1. 1Recipient data feeds a template with clearly marked personalisation slots
  2. 2A language model composes the copy from that data and a set of tone rules
  3. 3The generated message passes checks on length, format and forbidden phrasing
  4. 4Sending happens at a scheduled time with a daily cap
  5. 5Replies and delivery failures return to the sheet and halt further sends
1Recipient data from a sheet2Copy generated by the model3Length and format checks4Sending with a daily cap5A reply halts further sends
Flow: recipient data, copy generated, quality checks, capped sending, replies handled.

What I would refine for a client

I would add a hard human approval step before the first send to any new group. A model can write a sentence that sounds right and is untrue, and in customer communication that is not a risk worth taking automatically.

Do you have a similar process?

Describe it in a few sentences. I will tell you whether it can be automated and what it would cost — including when the answer is that it is not worth it.