Underperformance detection
Know what each system should have made today.
Sol AI by Modular Intelligence detects solar underperformance by learning what each system should produce in its own weather, then comparing real output against that every night. When a system falls short, the gap is flagged and priced in dollars a day.
The problem
Online and fault-free is not the same as producing.
A system can be online, report no faults and still make less than it should: a dead string, a failing optimizer, new shade. Comparing today with yesterday, or with a fixed number, misses it, because the weather changes every day.
Without a weather-adjusted baseline, every overcast day looks like a problem and every slow failure looks like weather.
What we do
How Sol AI measures it
- Within its first week on a new site, Sol AI learns what that site should make in its own weather
- New readings about every 11 min (fleet median)
- Every night, each site and each string the inverter reports is checked against what it should have made
- Each shortfall is priced in dollars a day, and the most expensive comes first
- After a repair, the case stays open until production shows the fix worked
AI agents on your team
The AI agents doing the work.
Onboarding agent
Starts watching a new site the day it's switched on.
The day a site comes online, the Onboarding agent starts reading it. By the end of the first week it knows what that site should make in its own weather.
Baseline ready — week onePerformance agent
Every night, checks every site and string against what it should have made.
Every night it compares every site against what it should have produced that day — down to each string (a row of panels wired together) where the inverter reports them. It prices each problem — “losing $28 a day” — and puts the most expensive first.
Losing $28 a dayCase agent
Owns each problem until the site proves it's producing again.
Each problem gets a case, and the Case agent keeps it open after the repair visit. It closes only when the site's own production shows the fix worked — “Fixed ✓ verified”.
Fixed ✓ verifiedCustomer agent
Answers the plant owner's questions on WhatsApp, 24/7.
It answers the owner's questions about their own system on WhatsApp, any hour, from live plant data, and sends alerts and reports on the same thread. Like an engineer on call for every plant, all day, every day.
Replies from live plant dataFAQ
Questions, answered
How do you detect solar underperformance?
Compare actual output with what the system should have produced in the same weather. Sol AI by Modular Intelligence learns each system's expected output in its first week, checks every site nightly against it, and flags and prices any shortfall.
What is a weather-adjusted baseline?
The output a solar system should be producing given the sunlight, temperature and cloud cover at its location. It is what separates a real fault from a cloudy afternoon.
How long before Sol AI knows what a new system should make?
About a week. The Onboarding agent starts reading a site the day it is switched on, and by the end of the first week it knows what that site should make in its own weather.
Which of your sites is quietly underperforming?
Bring a list of sites. On a 20-minute call we will show you what the comparison looks like on your own fleet.
Or email [email protected].