Week 17: Applications and Implications

Final Project: Smart Plant Buddy

Built by human hands, refined by AI conversations

Visual Identity Co-Creation

To kick off the project documentation, I wanted a header image that would make the Smart Plant Buddy immediately recognizable — not just a generic sensor render, but something with warmth and personality. I discussed this with the AI and proposed a Studio Ghibli-style visual to give it charm. The assistant agreed and generated the following image based on a shared vision:

Conversation Summary:
I wanted a graphic for the top of the page — either a sensor or a logo. I suggested something “like Ghibli style, comic-like.” The AI proposed a friendly, illustrated plant-sensor hybrid with anthropomorphic vibes and a soft palette. We both agreed it should feel whimsical but modern.
Final Prompt Used:
“This charming digital illustration presents the ‘Smart Plant Buddy’ in a Studio Ghibli-inspired design, highlighting a friendly, anthropomorphic device with a potted plant on top against soft sky-blue hues. The device's simplistic, inviting face is paired with earthy tones and soft textures to create a warm, hand-painted feel, with the bold ‘SMART PLANT BUDDY’ text prominently displayed below.”
Smart Plant Buddy logo illustration

1. What does it do?

The Smart Plant Buddy is a wireless, battery-powered plant monitoring device that senses key environmental parameters: soil moisture, temperature, humidity, and light levels. The data is uploaded via Wi-Fi and accessed through a mobile-friendly web interface. Bonus feature: an NFC tag lets users tap their phone to instantly access the dashboard. It’s designed to feel like a small product, not a prototype — and no, it doesn’t water your plant (yet), but it knows when your plant is being neglected.

Visualizing Functionality: Black Box Diagram

To explain the system's core functionality, I wanted a clear, minimalistic diagram. So I asked the AI to generate a "black box"-style schematic showing the inputs and outputs of the device, while keeping it visually clean for documentation. The following image was generated from that conversation:

Conversation Summary:
I mentioned that I wanted to include a black box diagram to explain the basic functionality of the Smart Plant Buddy. The goal was to show environmental sensor inputs, and outputs like Wi-Fi and a phone interface, without going into circuit-level detail. I said it should be high-contrast, clean, and work like a visual TL;DR of what the system does.
Final Prompt Used:
“This 2D digital diagram showcases how a plant monitoring system works via a 'black box' labeled 'SMART PLANT BUDDY,' broken into three sections: Inputs, the black box, and Outputs. The image highlights key elements such as soil moisture, temperature, humidity, light level, and connections to Wi-Fi and smartphone outputs, all in a bold, high-contrast black-and-white layout for clarity.”
Smart Plant Buddy black box schematic

2. Who’s done what beforehand?

Smart plant monitors are not a new idea — but execution, context, and design vary wildly. I reviewed several FabAcademy projects and some commercial implementations to understand what worked and what didn’t. Here's what I found:

Conversation Summary:
I researched several plant monitoring projects, including FabAcademy final projects and commercial systems. I shared the links with the AI and asked for a structured table comparing their goals, strengths, weaknesses, and what stood out in each. The goal was to situate my own design within the broader landscape and see what others had tackled — and what they hadn’t.
Final Prompt:
“I found some plant monitor projects from FabAcademy and other sources. Some are finished, some are just assignments. Here are the links. Can you make a table that compares each one — description, strengths, weaknesses, and notable features — so I can better position my own project?”
Project & Link Description Strengths Weaknesses Notable Features
Deepak Chaudhry (2021) Soil and air sensor box with solar panel and GSM communication. Good power independence, modular design. Data interface is limited, no documentation of enclosure build. Solar powered, focuses on agri-scale deployment.
Yesenhia Parra (2024) Hydroponic system with sensors and pH control, great housing design. Visually polished, nice 3D modeling, good integration. No interface for remote control or data logging. Clean physical design; focused more on structure than UI.
Elsa Cui (2025) Soil moisture + OLED plant monitor in a minimalist enclosure. Elegant documentation, strong UX thinking. Limited scope — only one sensor, no connectivity. Focus on physical form and aesthetics.
Liam Howell (2024) Plant pot with capacitive sensor and data logging to SD card. Complete integration, good fabrication process. No wireless data or dashboard; offline-only approach. 3D-printed pot design with embedded sensors.
Angela Mejía (2024) Plant monitor as part of application week — temp, humidity, soil. Good sensor choice, complete MVP. Not a final project; lacks integration and packaging. Simple and approachable prototype.
Elecrow Smart Plant Commercial product that senses moisture and alerts via app. Clean industrial design, app interface. Closed system; no customization, no transparency. Bluetooth alerts; sleek and consumer-ready.
RoboCraze DIY Kit Educational soil sensor + display kit with tutorial support. Great for beginners, low cost, open-source examples. Very basic; no connectivity, limited expansion. Uses DHT11 + soil sensor + LCD in a breadboard format.

3. What did I design?

I didn’t invent plant sensors. I’m not here to reinvent the moisture-measuring wheel — and I’m okay with that. This project isn’t about novelty, it’s about execution. What I designed is a system that works: a device that’s wireless, battery-powered, compact, low-maintenance, and clean enough to feel like a product — not just a prototype.

But here’s the twist: I didn’t design it alone. I designed it *in collaboration with AI*, using a structured prompt-response workflow that helped me think clearer, faster, and deeper. Some might say that's cheating. I say it's engineering. I used what was available — just like I’d use a multimeter, or a library, or a PCB mill.

Before FabAcademy, I had little experience with coding, circuit design, or embedded systems. Through the weeks — and a ridiculous number of questions — I not only built technical skill, but also learned how to enhance my own decision-making by collaborating with AI. I didn't just design a thing — I designed *how* I build things from now on.

Here’s what I designed (with AI in the loop):

Conversation Summary:
I wanted to visually represent what this project was really about: not just building something, but co-designing it with AI. The FabAcademy is about learning to make (almost) anything, and I pushed that into “co-making with AI.” So I asked for a comical illustration that captured a human working alongside an AI, blending creativity and fabrication. I specifically asked for a mix of comic style with strong FabAcademy and maker culture references.
Final Prompt:
“This vibrant digital illustration showcases the dynamic collaboration between a human and an AI in creating innovative projects, blending bold lines and bright hues in a style drawing influence from both manga and Western comics. On the left, a focused young man sketches designs while standing next to a 3D printer, while on the right, an AI with glowing blue eyes and a sleek body points towards a holographic display, symbolizing the merger of human creativity and artificial intelligence in design.”
Human Creativity

4. What materials and components were used?

This is the list of components used — or painfully ruled out — during the co-design of the Smart Plant Buddy. It’s organized by function, sourced mostly from UNIT Electronics in Mexico, and annotated with tears of experience.

Conversation Summary:
I told the AI what components I had, what I could source locally, and where I was stuck. We compared sensors (e.g., resistive vs. capacitive soil probes), debated 3.3V vs 5V issues, and decided to drop the RS485 sensor because — sanity. The list evolved based on GPIO limits, power supply design, physical constraints, and how likely things were to actually arrive on time.
Final Prompt:
“Let’s build a bill of materials for the Smart Plant Buddy. Group components by type: sensors, power, control, casing, and accessories. Use the UNIT Electronics website links where possible. Add price estimates in MXN and call out any potential compatibility quirks. This list is flexible — we’ll improve it if time or rage demands it.”

Sensors

Microcontroller & Power

Casing and PCB

Accessories & Miscellaneous

Total Estimated Cost

~ MXN $575 (~USD $34)
Cheaper than therapy, and now my plant is thriving.

5. Where did they come from?

All the electronic components — sensors, microcontroller, charging module, and battery — were sourced from UNIT Electronics, a solid Mexican supplier that had everything I needed (and some stuff I didn’t but almost impulse-bought anyway). The NFC sticker was ordered from Amazon, and all 3D printing filament, headers, wires, and fabrication materials were provided by the Fab Lab Puebla. Also, a big shoutout to the lab's “box of mystery parts” — truly the unsung hero of this build.

6. How much did they cost?

Here's the breakdown of the main components used in the Smart Plant Buddy, with rough prices in Mexican pesos (MXN):

Total Estimated Cost: ~MXN $575 (~USD $34)

Which is a pretty solid price for a fully wireless, sensor-packed, AI-co-designed botanical sidekick. And way cheaper than a month of plant therapy.

7. What parts and systems were made?

AI can’t physically build anything (yet), but I can — and did. With its support in decision-making, sanity-checking, and “don’t burn your lab” warnings, I made the following parts and systems for the Smart Plant Buddy:

AI didn’t touch a soldering iron or run the laser cutter — but it helped decide what was worth building, how to route connections, how to avoid frying components, and how to balance ambition with feasibility. Honestly, if it could hold a screwdriver, I’d be out of a job.

8. What processes were used?

This project is powered by sensors, solder, and a highly unstable relationship with multiple fabrication machines. Here's the breakdown of the actual processes used to bring the Smart Plant Buddy into the physical realm:

This wasn’t just about using machines — it was about learning *which* process made the most sense for each part, and not overcomplicating things that could be done simpler. That’s real design maturity. Also, everything smells faintly of burnt MDF now.

9. What questions were answered?

This project was driven by questions — not just technical ones, but practical, strategic, and existential ones. Every design decision came from a process of asking, testing, validating, and occasionally panicking. These weren't theoretical “nice-to-think-about” questions. They were real doubts that determined whether this project would actually work — or be another unfinished folder on my desktop.

Here are some of the actual questions I had — and how I answered them:

This wasn’t just a project of making — it was a project of figuring things out, one constraint at a time. And I didn’t figure it out alone. Every resolved doubt was a step forward, and every step made the final result feel possible.

10. How will it be evaluated?

This project will be evaluated based on how well it integrates the core FabAcademy skills into a complete, functional system — and whether it does what it claims to do without releasing smoke or tears during the final demo. Specifically:

In short, it will be evaluated as a functional, embedded system that merges physical fabrication, digital design, and network interaction. And if, during the demo, the plant looks happy and I look slightly less stressed than expected — that’s bonus points.

And yes — the AI part.

AI played a major role in this project. From selecting components to designing the wiring strategy, from helping write documentation to even suggesting how to frame this very section — ChatGPT was involved. But to be clear:

I know some people may give this side-eye — like using AI is a cheat code. But here’s the truth: this wasn’t an AI-built system. This was a human-built system with AI acting as a design partner, mentor, and occasionally a sarcastic sounding board. I still had to learn, decide, fail, rewire, and actually build the thing. The AI didn’t hand me a project — it helped me make mine better.

And if you're still unsure about how to evaluate that, well — just ask yourself: when Neo jumped off the building in the Matrix, did the simulation care if it was the Oracle’s idea or Morpheus’? No. It cared if he stuck the landing. So here I am, final project in hand. Let’s see if it lands.

11. What tasks have been completed?

The foundation of the project is fully defined: components selected, system architecture clarified, and the design direction locked in. A full review of sensor compatibility, power planning, and microcontroller limits has been completed through co-design with AI — reducing guesswork and giving the build phase a focused trajectory.

Actual physical fabrication begins this week— including sensor testing, enclosure design in Fusion 360, and PCB prototyping. This section will be updated as things go from “on paper” to “on table.”

12. What tasks remain?

The final integration phase is just around the corner. Now that the design and planning groundwork is set, the remaining tasks revolve around physically building, programming, testing, and—most importantly—making everything work together in a coherent, compact form.

In other words: everything that turns a well-scoped idea into an actual working system. This section will be updated throughout the build process (and probably rewritten under stress).

13. What has worked? What hasn’t?

The planning phase worked surprisingly well — probably because it was a collaborative effort between human intuition, late-night stress, and a very patient AI assistant. Here’s a breakdown of what’s already proven solid, and what still lives in the grey area between “we’ll figure it out” and “please don’t short circuit.”

This section will evolve (read: fill with battle scars) once the fabrication starts. For now, what’s working is the structure and mindset — and what’s not working is mostly still waiting its turn.

14. What questions need to be resolved?

Even with solid planning and a lot of simulated confidence, several key questions remain open — questions that only hands-on testing, real-world messiness, and probably some debugging at midnight can answer. These are the remaining unknowns:

These questions aren’t blockers — they’re just the final boss fights waiting at the end of the level. I’ll be updating this section as things get built, tested, and either confirmed or violently disproven.

8. What will happen when?

The plan is simple: four days, 15-hour shifts, questionable nutrition, and one final demo standing between me and FabAcademy graduation. Everything begins on Monday June 10 — and by Wednesday June 12, it needs to be built, working, and looking fabulous. No pressure.

Every day is a full sprint. Failure is not an option — just a learning opportunity I don’t have time for.

15. What have you learned?

I’ve learned to solder, to wire, to code, to debug, to make things break and then pretend that was part of the plan. I’ve learned how to talk to machines, how to talk to AI, and most importantly — how to talk myself out of giving up.

I’ve learned that a good question is more powerful than a perfect answer. That progress isn’t a straight line, and that Google, ChatGPT, and sheer stubbornness can build almost anything.

But most of all, I’ve learned that the real skill isn’t electronics or CAD or Python — it’s turning confusion into momentum. That if you give me a problem, a microcontroller, and five hours of panic... I’ll come back with something that works. Barely. But it works.

So yeah — I’ve learned a lot. And if I could do all this in a few months, with minimal prior experience and a lot of weird late-night decisions… imagine what comes next.

Mic dropped.

Back to Final Project | Back to Top