Lunatechs Event · May 30, 2026
A cafe session for trying DeepSeek with OpenCode.
DeepSeek supplied the model. OpenCode supplied the coding-agent interface. People brought laptops and tried to make something run.
I helped host as one of the Lunatechs organizers. I spent most of the afternoon on setup questions, table moves, and helping beginners get unstuck.
Drop By Dough helped because a cafe makes the first question easier to ask. People could grab food, move seats, and come back to the terminal.
Hosting the room
The first step was still the hard part.
Many AI coding events assume everyone is already comfortable with terminals, installs, model keys, and broken agent output. Real rooms are not like that.
My job was simple: help first-timers ask the first question, then leave stronger builders alone long enough to test the tool.
The loop
Pick a small task, run it, inspect the diff.
Choose the task
Pick something small enough for the session, but real enough to touch files in a project.
Let the agent propose
Use OpenCode for the first pass, then read the commands and file changes before accepting them.
Run the thing
A working screen, a failing command, or a broken import tells you more than another tool comparison.
Keep receipts
Do not accept a big patch blind. Read the diff, rerun the command, and keep the task small.
The session
Free credits removed one boring blocker.
Alibaba Cloud joined as tech partner, gave a short database-tech session, and helped people start with DeepSeek credits.
Once people had credits and a working setup, the questions got better: what should I build, what did it change, and how do I know it worked?
What we actually learned
The rough edges showed up quickly.
Cost, model choice, API setup, and review habits all matter once you stop talking about AI coding and start running it.
Some parts felt smooth. Some parts broke. That was the point of testing it in a cafe with other people nearby.
Cafe setup
Donuts, laptops, and people asking basic questions out loud.
The venue mattered. A cafe makes it normal to move around, ask someone at the next table, and come back to the screen.
That helped the beginners. Nobody had to pretend they already knew the terminal commands.
Closing note
The stack is still messy enough to test with other people around.
Run the tool. Watch it break. Compare notes before pretending you have a strong opinion.
With thanks to
Partners who helped with the practical parts.
Venue partner
Drop By Dough
Opened the Central space and kept the afternoon closer to a cafe work session than a classroom.
Tech partner
Alibaba Cloud
Shared a database-tech session and helped people start with DeepSeek credits.
