AI-assisted coding, for a complete novice in development like me, is a revolution. It allows me to turn ideas into reality in a few hours where it would have been impossible before. But this speed comes at a price we too often overlook: technical debt.

The Development Context

Recently, as I mentioned here, I developed a small web application named HYCU Upgrade Path. The concept is simple: a clean interface to generate and visualize the upgrade path from version A to version B of backup software.

To do this, I used Lovable, a “no-code/low-code” solution that allowed me to get the application online in a few days.

A few weeks later, I decided to rebuild the application using Gemini, Google’s multimodal IA, to see what it was capable of.

I gave similar instructions to both AIs. I expected a functional and similar result from both sides. What I didn’t foresee was the abysmal gap in the code structure… And the time spent.

We’re talking about a difference of 628kb versus 115kb. We’re talking about going from 79 files to 21 files. We’re talking about one week versus half a day. A chasm.

The Lovable Experience

On paper, Lovable sells a dream: you describe, it displays. And it’s true, the user experience is stunningly fluid. I described my need for HYCU Upgrade Path, and in a few iterations, the application was running before my eyes. Visually, it was clean with a few adjustments needed. Functionally, it was a bit more shaky.

I’m not a developer, but my first reflex was to check the generated repository. And there… it was a rude awakening.

For an application as simple as an upgrade path generator, Lovable spawned a veritable Rube Goldberg machine:

• 79 files created.
• 8 nested folders.
• A total weight of 628kb.

Lovable fragmented the application into a myriad of tiny atomic components, generated duplicate configuration files, and imported libraries “just in case”. Looking at the components/ui folder, I found 38 files! Accordions, badges, carousels, “toasters”… when my app uses only a fraction of all that.

This is the “Black Box” approach: as long as it works on the surface, no one cares about the elegance of the engine. The result is a heavy application that is difficult to maintain over time. And for me, that’s a first red flag.

Regarding code modification through iteration, I didn’t find it very precise. I was forced to move forward in very small steps, which lengthened the development time… An iteration a bit too bold? It broke my interface half the time or gave a result that wasn’t what I expected. In short, not hyper-satisfied with the experience, although I managed to reach my goal in a short week of discussions.

The Gemini Experience: The Software Architect

After this week of struggle and a few weeks in production, I tried the same experiment with Gemini. Same prompt, same need, just “to see” what Google’s AI was capable of.

The result? At first, I thought pieces were missing.

• 21 files in total.
• 5 folders.
• 115kb on the scale.
• Development time: half a day.

Gemini adopted a radically different, much more mature approach. Instead of importing a massive graphic library, it structured the code and kept only what was truly necessary.

Opening the src/components folder, the difference is glaring: 7 files, all essential. No noise, no useless components.

Even more impressive, Gemini provided me with a “Production Ready” architecture. Where Lovable gave me a messy frontend, Gemini cleanly separated the frontend from the backend, and even generated a docker-compose.yml and a deployment script (deploy.sh) for my VPS. It didn’t just code the interface; it thought about how I would run it, complete with a nice README.md.

The Technical Match: Obesity vs Efficiency

Let’s compare apples to apples. Here is the scorecard for the same functionality:

Metric	Lovable (The Factory)	Gemini (The Architect)	Gain
Project Weight	628 kb	115 kb	-82%
File Count	79	21	-73%
Philosophy	Full UI Kit (“just in case”)	Clean, readable code	Easier Maintenance
Architecture	Monolithic Frontend	Dockerized (Front + Back)	Ready to Deploy
Dev Time	1 week	1/2 day	Productivity x10

Why “cleanliness” matters (even for non-devs like me)

You might tell me: “We don’t care about the code or the number of files as long as the app works!”. That is a mistake.

Maintenance: Manually modifying the project created under Lovable is like looking for a needle in a haystack of 79 files. With Gemini, every file has a name that describes actually what it does.

Consumption: 628kb of code to load for a mobile user is heavy. 115kb is almost instant.

Cognitive Load: When AI generates “messy” code, it has a harder time reading itself for subsequent iterations. That’s why Lovable “broke” the interface: it got lost in its own complexity.

Conclusion: The Choice of Tool Makes the Difference

My verdict is clear.

If you want to prototype an idea in 10 minutes to show your boss that “it moves”, Lovable creates a good illusion. It’s spectacular, it’s visual, it’s a beautiful tech demo.

But if you want to build a lasting application that you can evolve without breaking everything, Gemini is far more efficient in my eyes.

The ecosystem is evolving fast. I haven’t tested it yet, but a tool like Claude Code is probably capable of doing even better.

October 2023. The situation is tense: the price per kWh is skyrocketing, and my newsfeed is flooded with ads for solar panels promising “total autonomy” or “free energy.” Being naturally suspicious (and a bit of a geek), I took out my Excel spreadsheets before taking out my checkbook. I invested €13,900 for 6kWp of power. Two years later, with 17.4 MWh produced, was it worth it? Spoiler: real-life figures beat my simulations, but the devil is in the details.

1. The Genesis: Why Turn My Roof into a Power Plant?

You don’t wake up one morning deciding to drop nearly 14,000 euros (before subsidies) just to please the planet. It is a calculation. My goal was twofold: to secure part of my energy costs for the next 20 years and, let’s admit it, the technical pleasure of managing my own production.

The Context: Buying Electricity in Advance

For the neophyte, seeing this as an expense is a mistake. It is a pre-purchase. By installing panels, I decided to buy a stock of electricity at a fixed price (the cost of installation divided by future production) rather than renting this energy from a supplier whose rates are indexed to geopolitical crises I cannot control.

But be careful, for this calculation to work, you shouldn’t size the installation based on guesswork.

Consumption Analysis: The Essential Prerequisite

Before even contacting an installer, I audited my own home. Many make the mistake of looking at their annual global bill. That’s insufficient.

You need to understand when you consume.

Solar only produces during the day (no kidding). If 80% of your consumption happens at night (electric heating without inertia, nightlife), solar without batteries will be a financial failure.

I extracted my hourly data via the Enedis website (thanks to the Linky smart meter) to isolate my energy “background noise,” what we call the baseload.

This is the house’s incompressible consumption when “nothing” is on: fridge, internet box, VMC, standby devices. For me, this baseload justified a production base, but to reach profitability on 6kWp, I had to be able to shift my heavy consumers (Washing machine, Dishwasher, Water heater) to the daytime. It was this “load shifting” potential that validated the project.

2. The Technical Study: Choosing Without Getting Scammed

Once the need was validated, I had to choose the hardware. The solar market is a jungle where passionate artisans rub shoulders with eco-scammers. Here are my technical choices and, above all, why I made them.

Self-Consumption with Surplus Sale: The Logical Choice

I opted for the standard model: I consume what I produce first, and what I don’t consume is automatically injected into the grid and sold to EDF OA (Purchase Obligation).

In October 2023, this contract guaranteed a fixed feed-in tariff (around 13 cts/kWh) for 20 years. This is a major financial security that allows amortizing the installation even if I am not home to consume.

The Hardware: DualSun and Enphase, the French-American Couple

For my 6 kWp installation, I selected:

• 16 DualSun FLASH 375 Half-Cut Panels (Total: 6000 Wp)
• 16 Enphase IQ8M Micro-inverters
• One Enphase Envoy S-Metered communication gateway

Why this choice?

1. DualSun Panels (375 Wp): It’s a French brand (manufactured in Asia, let’s be honest, but French engineering). The “Half-Cut” technology (half-cells) allows better management of partial shading and reduces resistive losses. They are robust and aesthetically sober (black frame).
2. Micro-inverters vs Central Inverter: This was the big debate. I chose Enphase IQ8M micro-inverters.

Why the IQ8M?

Unlike a central inverter (like SMA or Fronius) which manages the entire string in series (if one panel fails or is shaded, the whole string drops), the micro-inverter manages each panel independently.

But why the IQ8M model? It is Enphase’s latest generation capable of creating a micro-grid (although I don’t use the “Sunlight Backup” mode without battery yet). The “M” suffix indicates an output power adapted to my 375Wp panels. With a peak output power of 330VA, the DC/AC ratio is 1.13, which is excellent for avoiding clipping while maximizing production in low light.

The “Plug & Play Kits” Parenthesis and the Efficiency Trap

Before signing my quote, I obviously looked at the “Plug & Play” kits found in DIY stores. On paper, it’s seductive: no craftsman, plug it into a socket, and you’re good to go. But for 6kWp, this solution was not viable, and one must warn against a frequent marketing mirage.

A 400W panel will never produce 400W if it is poorly oriented. Balcony kits, often placed vertically (90°) or with approximate tilt, lose a huge amount of efficiency compared to an optimized roof installation (generally 30-35°).

The trap is confusing Peak Power (what the panel can output in a lab) and Real Production (useful energy).

On many kits, the inverter is deliberately undersized (to respect the injection limit on a simple socket). You buy a 420Wp panel, but the micro-inverter caps at 350VA. This is called clipping. It’s not serious in itself, but it’s a net loss in the middle of summer. For my 6kWp project, I wanted total coherence between the capacity of the DualSun panels and that of the IQ8Ms to milk every available photon.

3. Installation and Commissioning (October 2023)

Once the hardware was validated, time for action. The installation took place in late October 2023.

Installing 16 panels is not trivial. You have to manage the layout on the roof, the routing of DC cables under the tiles, and the run down to the electrical panel. The advantage of Enphase micro-inverters here is safety: you don’t bring high-voltage DC current (dangerous in case of electric arc) down into the house, but directly 230V AC current.

On the administrative side, do not underestimate the delays. Between the prior declaration at the town hall (DP), the grid connection request to Enedis, and the Consuel inspection (mandatory to validate electrical safety before injecting), it is a journey that requires patience. In my case, everything was wrapped up for effective commissioning at the end of 2023.

4. Production and Monitoring: The Truth of Numbers (2024-2025)

This is where the geek takes over. After more than two years of perspective, I can look up from theoretical estimates to give you the reality of the field.

Monitoring Tools: Enphase Enlighten

To manage it all, I use the Envoy S-Metered gateway. Note the “Metered.” Unlike the standard version which only measures production, this one uses measurement toroids (current clamps) placed on the house’s main supply.

Result: I see what I produce, but more importantly what I consume and what I import/export in real-time.

Without this visibility, self-consumption is done blindly.

Gross Production: 17.4 MWh in Two Years!

Here is the data extracted from my tracking for an installed power of 6 kWp:

Year	Total Production	Performance Ratio (kWh/kWp)
2024	8.9 MWh	~1,483 kWh/kWp
2025	8.5 MWh	~1,416 kWh/kWp

Data Analysis:

1. Weather Variability: There is a production drop of about 4.5% between 2024 and 2025. This is normal. The sun is not an absolute constant from one year to another.
2. Redoubtable Efficiency: With a ratio approaching 1,500 kWh produced per kWp installed in 2024, my installation performs extremely well (the national average is often between 1100 and 1300 depending on the region). The DualSun + Enphase combination works wonders, aided by near-perfect orientation/tilt (33% tilt for a near due south orientation) and correct panel ventilation (which lose efficiency when they heat up too much).

Self-Consumption: The Sinews of War

Producing is good. Consuming is better (financially).

• Self-consumption rate 2024: 46%
• Self-consumption rate 2025: 44%

Concretely, this means that I directly consume about 45% of my production. The rest (55%) comes from the grid.

Despite my efforts (delayed start of machines, water heater during the day), I stagnate below the 50% mark. Why? Because in summer, days are long and production explodes (sometimes 40 kWh/day), far beyond the house’s needs. Without a physical battery or an electric vehicle to charge on the weekend, it is difficult to go higher on a 6kWp installation. This is where selling the surplus becomes vital for profitability.

5. The Time for Assessment: Profitability and Real ROI

Let’s talk cash. We hear everything and anything about solar profitability. Here are my real figures, bill in hand, unfiltered.

The Final Cost of the Operation

For a turnkey 6 kWp installation (hardware + labor + procedures), the bill amounted to:

• Initial Investment: €13,900 incl. VAT
• Self-consumption Bonus (State): – €2,000
• FINAL REAL COST: €11,900

Scenario 1: My Reality (2023 Contract)

With a self-consumption rate of about 45% (the remaining 55% coming from the grid) and a surplus feed-in tariff fixed at €0.13/kWh (the rate in force at the time of my request), here is my average annual yield (based on an average of 8.7 MWh/year):

1. Bill Savings (Self-consumption):~3,915 kWh that I did not buy from EDF (base €0.25/kWh) = €978 in savings.
2. Surplus Sale (Injection):~4,785 kWh sold to EDF OA (€0.13/kWh) = €622 in income.

Total Annual Gain: ~€1,600

Return on Investment (ROI): €11,900 / €1,600 = 7.4 years.

Verdict: Amortization in less than 7 and a half years for hardware guaranteed for 20 or 25 years is an unbeatable financial investment, far superior to any savings account.

Scenario 2: If I Had to Do It Again Today (The 4-Cent Trap)

This is where my article should serve as a warning. The rules of the game have changed. Recently, the surplus feed-in tariff dropped drastically to around €0.04/kWh (depending on the quarter). Let’s redo the calculation with this new parameter, keeping the same installation:

1. Bill Savings: €978 (Unchanged).
2. Surplus Sale (New Tariff):~4,785 kWh * €0.04 = €191 (instead of €622!).

Total Annual Gain: ~€1,169

New ROI: €11,900 / €1,169 = 10.2 years.

This drop in the feed-in tariff upsets the strategy.

• In 2023 (my case): Selling surplus was a pillar of profitability. I could afford to inject almost 60% of my production without too much pain.
• Today: Selling at 4 cents covers almost nothing. The absolute priority is no longer to produce a lot, but to consume everything. This highlights the interest in virtual or physical batteries, which were economically unviable two years ago but, with such a low buyback rate, are becoming an option to seriously study to avoid “wasting” 60% of one’s production.

6. Conclusion

After more than two years, do I regret my €11,900? Absolutely not.

Producing 17.4 MWh of green energy from my roof is a daily satisfaction. Seeing my Linky meter display “0 VA” of consumption while the oven and washing machine are running is a pleasure one never tires of.

Technically, the DualSun / Enphase couple is of exemplary stability: no breakdowns, precise monitoring, production on target.

However, if you start today, do not blindly copy my economic model. Do your calculations with the current feed-in tariff. If you cannot shift your consumption to the daytime, the ROI risks drifting away. Solar remains profitable, but it now requires being even smarter about its management.