Dev Update #1: MVP works. Now I want to make it easier to build.

[jamro]

The first version of Tiny Software Engineer is alive, and I’ve reached a fun milestone: I’m now using the robot while developing the robot itself. 🙂

It sits on my desk, connected to Cursor, reacting to what my coding agent is doing while I work on its own code.

What works today

V1 already supports a growing set of animatronic behaviors:

• 👋 Welcome
• ⌨️ Typing
• 📖 Reading
• 🤔 Thinking
• 🔔 Bell / job done
• ☀️ Wake up
• 🌙 Go to sleep
• 👀 Attention needed
• ⚠️ Error
• 🛑 Task aborted
• 💀 Robot dead

Not all of these are integrated with Cursor yet, but all of them can already be triggered through the robot’s API and testing UI.

Configuration is also handled through a web UI, so changing the robot’s settings doesn’t require recompiling or modifying firmware.

Integrations

Out of the box, Tiny Software Engineer currently integrates with Cursor.

But the robot itself isn’t tied to Cursor. Its behaviors are exposed through an API, so integrations with other coding agents, IDEs, automation tools, or completely different use cases can be built on top of it.

Something interesting happened after sharing the project: I received some really useful feedback after posting the project on Reddit and LinkedIn. A recurring theme was:

“I’m a software engineer. I’d love to build one, but I don’t own a 3D printer / I’ve never worked with electronics / I don’t know where to start with hardware”

That made me realize that the biggest barrier may not be the software at all.

So I created a dedicated Hardware for Software Engineers guide, explaining the hardware side from the perspective of someone comfortable with code but new to electronics and physical builds:

Hardware for Software Engineers guide

🎯 What’s next: make the robot easier to build

This is going to be the main direction for the next iteration. The first step is standardizing the components and making sure the recommended parts are easy to source internationally. I’m currently considering a few hardware changes.

Standardized servos

V1 uses PowerHD HD-1370A servos.

I’m considering switching the reference design to the Pololu Feetech FS0307 and adjusting the 3D model around it.

The goal is to establish a component that can become part of a reproducible reference build and is easily available

Standardized display

I’m also considering replacing the current OLED with the Waveshare 0.91” OLED Module and adapting the enclosure accordingly.

Again, the goal is predictable dimensions, wiring, documentation, and availability.

🖨️ No 3D printer required

I also want Tiny Software Engineer to be buildable by someone who doesn’t own a 3D printer.

I’ll prepare the 3D models for external printing services such as PCBWay and order a complete set myself to test the whole process.

Ideally, someone should eventually be able to download the files, send them to a printing service, order the documented electronics, and assemble the robot without having to redesign anything.

🔌 Simplifying the electronics

Another area I want to explore is reducing the amount of wiring and electronics knowledge required. I’ll experiment with simplifying the current electronics and, potentially, designing a dedicated PCB for Tiny Software Engineer.

That could turn a collection of modules and wires into something much closer to:
plug in the servos → connect the display → power it → configure it in the browser.

That’s the direction I’d like to move toward.

⸻

V1 proved that the idea works.

The next challenge is making it reproducible, especially for software engineers who have never built hardware before.

If you’re interested in building one, I’d particularly love to hear what would stop you today: 3D printing, soldering, sourcing components, assembly, cost, or something else?