The Sun Doesn't Come With a Dashboard.
Solar panels have one job, but when the inverter numbers drop, the cause isn’t always clear. SuryaWatch is our Environmental Hacks 2026 prototype that helps Indian homes plan rooftop solar and work out whether a dip in output is smog, dust or a real fault, with just a phone camera. It compares what your panels made against what the sunlight allowed, then tells you whether it was haze, dust or something else. Its most important feature might be knowing when not to clean, saving water and effort on smog days.
Solar panels have exactly one job: turn sunshine into electricity. So when the number on your inverter suddenly drops, the reason is rarely obvious. Is the sky hazy? Is there a layer of dust on the glass? Or is something actually broken?
Those are three very different problems, and each one calls for a different response.
HAZE / SMOGLook up before you wash. The weather may explain a broad dip.
DUSTClean when it makes sense. Estimate the loss before spending water and effort.
POSSIBLE FAULTA drop that's specific to your roof deserves a closer look.
Our starting point wasn't "every panel loses the same amount." Impact varies by place, weather, installation and upkeep. It was simpler: can we help people ask a better question when performance changes? That question became SuryaWatch.
First, plan for the roof you actually have
A solar estimate can feel like a spreadsheet wearing a hard hat. We wanted the first step to feel friendlier: find your home, outline the roof, enter your monthly electricity use (or just photograph your bill), and see where you land.
Recommended system
3.3 kW
About 6 panels of 550 W, making 4,999 units a year. Sized by your electricity use.
You pay after subsidy**₹1.13 lakh**
Benefit per month**₹2,509**
Pays back in3.8 years
System cost (3.3 kW × ₹60,000)₹1,98,000
PM Surya Ghar subsidy− ₹78,000
Delhi Solar Policy subsidy− ₹6,600
Example result from the demo. Illustrative estimates, not a quote or a subsidy guarantee.
Why show everything together? Because "save money with solar" isn't enough. People need to see the trade-offs and assumptions before they know what to ask an installer.
Then, give the inverter a translator
Once panels go up, the inverter display becomes a tiny scoreboard for the whole system. But the numbers are fiddly to read on a small two-line screen, and a number by itself rarely tells the story.
So the habit we designed is deliberately lightweight: take a photo of the inverter display, check what the app read, save it. Behind that, Amazon Bedrock turns the picture into structured readings, and treats unclear digits cautiously instead of pretending every photo is perfect.
SuryaWatch then compares what your panels made with what that day's sunlight should have allowed, and offers possible explanations.
⇩ Clean panels
Dust on panels: clean them
Leaving the haze aside, your panels made 19% less than the sunlight allowed. That's about ₹112 a week. Clean early morning or evening with plain water and a soft cloth.
Made10.9 kWh
Sunlight allowed13.74 kWh
Performance79%
A sample verdict from the prototype. It's an estimate, not a field-validated diagnosis.
We wanted the interface to help people investigate, not to cosplay as an omniscient solar engineer.
That wording matters. A low output figure is a clue, not proof. Weather, shading, inverter behaviour, system design and data quality can all muddy the picture.
The most important feature might be knowing when not to clean
It's tempting to treat every dip as a dirty-panel emergency. But washing on a hazy day won't fix a hazy sky. And if every rooftop nearby is dipping too, the cause is probably area-wide, not unique to you.
≡ Smog day
Smog day: your panels are fine
Haze (AOD 1.10, PM2.5 160) cut about 18% of the sunlight today. Your panels made what the hazy sky allowed. Nothing to fix.
Made11.6 kWh
Sunlight allowed12.3 kWh
Performance94%
On a day like this, a wash is wasted water and effort.
⌂ Area-wide dip
Low across your area: your panels are fine
Most SuryaWatch rooftops within 5 km were low today too (3 roofs, middle value 69%), so the sky cut the output, not dirt or a fault.
Made6.4 kWh
Sunlight allowed12.8 kWh
Performance50%
The neighbourhood check shares only a count and a middle value, never names, places or any single roof's figure.
The prototype also includes a short forecast, an estimate of dust build-up, and a cleaning suggestion that accounts for likely rain. In plain English: weigh the possible loss against the cost of cleaning, and see if rain will do the job for free.
It's a small design choice with a bigger idea behind it. A useful environmental tool shouldn't only tell people to do more. Sometimes it should help them avoid unnecessary action.
Make the next conversation easier
If SuryaWatch flags something, the next step shouldn't be "good luck explaining this over the phone." So we added an installer report that gathers readings, gaps and possible issues into something printable or shareable, plus a concept for evening email alerts.
We also designed for everyday access. The interface works in English and Hindi, and the core interaction is a phone camera, not extra monitoring hardware. That doesn't remove every barrier, but it lowers the amount of new equipment someone needs just to try the idea. The Impact page shows only aggregated totals, with no names and no precise locations.
Under the hood: a small team, a big cloud
SuryaWatch is a web-first prototype with AWS doing the heavy lifting: API Gateway and Python Lambda functions for the logic, DynamoDB for readings and results, S3 for inverter photos, and Bedrock for reading photos and bills. Weather and sunlight come from Open-Meteo and NASA POWER, with scheduled checks and SNS email alerts in the design.
Phone camera*→Browser (Amplify)→API Gateway + Lambda→DynamoDB · S3→Bedrock→*EventBridge + SNS
Built with AI assistance. Directed by people.
A hackathon is rarely a straight line. It's questions, trade-offs, deadlines, half-working bits and the occasional "wait, what if we did it this way?" SuryaWatch was team-driven and AI-assisted: we shaped the problem, chose the workflows, discussed features, coordinated the work and iterated. Claude helped speed up coding and drafting, but the product decisions, and the responsibility for what we present, are ours. Agrim also got hands-on with the AWS deployment, packaging and publishing the app rather than just putting "cloud" on a slide.
What this demo is, and isn't
The screens use sample data. Financial figures are estimates based on assumptions, not guaranteed savings or an installer quote. The haze, dust and fault verdict is a prototype heuristic, not a model validated against field measurements. Photo reading and scheduled alerts depend on the relevant services, permissions and data being available.
What we'd love next is testing with real rooftops, installers and users: how reliably inverter photos can be read, whether the explanations match conditions on the ground, how well the cleaning advice holds up, and where the interface still confuses people.
The real goal isn't a clever dashboard. It's fewer guesses.
A homeowner shouldn't have to wonder whether to wash a panel, ignore a warning or call an installer with nothing useful to say. If SuryaWatch makes that next decision a little clearer, we're heading the right way.
SuryaWatch · Plan it. Watch it. Every rooftop, every day.
Aryan Nirala · Agrim Chaturvedi · Prakhar Srivastava · Anshu Dhaka
Environmental Hacks 2026 · Track 03: Waste and Energy
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article