← Back to archive
Issue #1 · 2025-12-15 · 4 min read

Before you automate, understand the work

Automation fails when it copies confusion. This is my process for studying the human workflow before turning it into a bot, API, or worker.

Most business automation projects fail not because the code broke, but because someone automated a broken human process.

When a founder or operations manager says, "We need a bot to handle customer intake," my first question is never about the API, the webhook, or the database. It is always:

What does a human actually do when an inquiry comes in today?

The three rules before writing code

  1. Watch the work happen in real life. People explain what they think they do, not what they actually do. They skip edge cases, informal Slack messages, and manual overrides they do without thinking. If you automate their spoken explanation, the bot breaks on day one.

  2. Standardize the data before connecting pipes. If one person writes dates as DD/MM/YYYY and another writes Month DD, an automation script will choke. Define the contract first. The bot should enforce rules, not guess intent.

  3. Keep a human escalation path open. Every good bot knows its limits. The moment a customer asks something outside the workflow, route it immediately to an actual person with full context. Never let a bot reply with a dead end.

Automation is not about replacing judgment. It is about eliminating the repetitive, clerical tasks that distract people from exercising judgment.

Engineering & Automation Notes

Bi-weekly deep-dives into backend architectures and practical business automation teardowns.

Samuel Ekunyan

Samuel Ekunyan

Backend & Automation Engineer

Book a free Automation Audit →