The Beginning of Repetitive Template Work
In trade business, messages exchanged with buyers always follow a similar structure. Greetings, product introduction, technical specifications, price inquiry forms. Every week when reaching out to new buyers, we had to find the previous template, copy it, and change only the product name and specifications.
"Can we automate this part?" A team member's question became the starting point.
Giving Claude Code Simple Instructions
We gave simple instructions to Claude Code on our Windows device.
When entering a buyer name and product category in Google Sheets,
can you automatically generate the corresponding communication template in a new cell?
The basic structure should be: (1) Greeting, (2) Product description, (3) Technical specs, (4) Inquiry form
Claude Code proposed a small script using the Google Sheets API. The logic wasn't complex.
Results Beyond Expectations
In the first test, templates were automatically generated just by entering the buyer name and product. But a few days later, something more interesting happened.
Claude Code started detecting repeated message patterns sent to the same buyer and learned the tone and length from previous messages, reflecting them in the new templates. The familiarity level of greetings, the technical depth of product descriptions, and even the number of paragraphs in each section were being reflected.
"Oh, this buyer prefers short, clear messages." This kind of judgment was happening automatically.
Small Changes in the Team
For the first two weeks, we were busy validating template quality. We read templates generated by Claude Code, occasionally edited phrasing, and added specific buyer preferences.
But by the third week, an interesting pattern emerged.
Once template generation was automated, team members actually had more time to communicate with more buyers. They could shift their attention from message writing to "what content strategy should be included."
Also, when we compiled the auto-generated templates, we discovered patterns in communication styles by buyer. Some regional buyers need detailed technical information, while others only care about price and delivery time. These insights naturally emerged.
Next Steps
Now we're testing a feature where Claude Code automatically suggests the optimal message timing and frequency for each buyer. It will simultaneously conduct analysis, like "This buyer responds fastest to messages sent Monday morning."
This wasn't building a large system, but when a small repetitive task was automated, the team's rhythm changed. This was the most unexpected harvest from this experiment.