AI in Solar: How artificial intelligence is influencing the next wave of solar systems and businesses

Solar companies, manufacturers and service providers explain how AI is integrated into their products, and how AI is changing the way solar + storage projects are modeled, sold, operated, and maintained

Artificial intelligence may still conjure futuristic imagery for some in the solar sector, but for many companies, AI is already here — an integrated part of how projects are modeled, sold, operated, and maintained.

To cut through the buzzwords and get a grounded perspective, we asked a range of manufacturers, software providers, and service partners across the solar value chain how they are using AI in their business and product offering, and what they think the future looks like for AI in solar.

Their answers paint a picture of cautious optimism, real-world utility, and an industry still figuring out where AI adds the most value. Some companies are already deploying AI to automate design and permitting workflows. Others are using machine learning models to predict energy yields or system degradation. Several emphasized AI’s power in streamlining customer experiences — whether through faster proposals, more accurate forecasting, or smarter service dispatch.

But, there are warnings too. Several respondents called out overpromised “predictive” tools that can’t yet deliver, or the misconception that AI is a magic fix for poor data quality. Many stressed the importance of human oversight and the foundational need for clean, structured data to get meaningful results from any AI implementation.

So, how are companies using AI in their platform, product, or services? Here are the standout responses in each product category.

Subscribe!
This feature originally published in the Q3 2025 issue of Solar Builder magazine. Read the whole issue and subscribe here.

Design and sales proposals

Aurora Solar continues to invest in the effectiveness of its AI and automation products. Similar to a human designer, Aurora AI produces higher-quality designs when higher-quality data is available, and so it leverages the best satellite and LIDAR data for each project location. Most recently, Aurora has increased the training data for Aurora AI by a factor of four, from approximately 500,000 to 2 million designs. The increase in training data and multiple updates and improvements to the model infrastructure have yielded a significant increase in accuracy and decrease in run times.

“Our Aurora AI grew by over 35% from last year and now generates 3D models in under 10 seconds (down from 30 seconds last year),” says Curtis Merring, staff product marketing manager at Aurora Solar.

OpenSolar launched AI Auto System Design in February 2025, initially to ~750 dealers in the Sunnova network. This month, they unveiled OpenSolar 3.0, a next-generation platform for solar installers globally, that combines AI-powered design, lead generation, and a fully integrated workflow — all free of charge. “Our AI Auto System Design capability reduces time-to-proposal from several minutes for the average user to under one minute, including multiple system options,” says Andrew McGuigan, general manager, Americas, at OpenSolar. It minimizes design errors and rework and has flexible default parameters including max offset, max fit, max savings.”

Mounting manufacturer Pegasus introduced a new AI-powered feature to help people complete permit pack request forms more easily. By toggling on Smart Assist, users see helpful hints for each field of the request form.

“Smart Assist can save half the time from inputting basic information when requesting permit packs,” estimates Kai Stephan, CEO and founder of Pegasus. Their second AI addition involves full project bill of material (BOM) logic, which eliminates the design-tool-to-Excel-to-emailed-purchase-order-PDF process by integrating all those steps right in Glide. “Imagine your team can upload a site survey or sketch of an array, and then Glide AI instantly generates an optimized structural layout, compiles and submits a permit pack aligned to local AHJ particulars, generates a full project BOM based on your team’s typical usage, and coordinates with local distribution for pick/pack/ship. That’s what’s coming with Glide.”

Enerflo deploys AI to streamline the entire solar sales-to-installation process. Features such as QuickQuote, Title Checks, Smart Pulls, and AI-driven solar designs and proposals expedite the front-end sales process. A standout feature is Enerflo’s integration with Aerialytic, which enables sales representatives to generate AI-driven solar designs and interactive proposals in under 2 minutes. Taken together, according to a case study by Ohm Analytics, installers using Enerflo’s platform have achieved a 62% reduction in project timelines while nearly doubling their project volume.

“This rapid turnaround is achieved through high-resolution aerial imagery and 3D modeling, which calculate factors such as roof pitch, shading, and optimal panel placement,” says Rebecca Taylor, head of marketing. “The result is a significant reduction in customer acquisition costs and a more efficient sales process.”

Enerflo also has AI-powered tools like the Battery Sizing Tool and Savings Calculator & Forecaster to provide personalized energy consumption analyses and savings projections that factor in state-specific net metering policies and utility rates.

On the flipside, Chetan Chaudhari, CEO with PowerUQ, says they intentionally do not use AI in their core computational engine for larger-scale project production modeling. PowerUQ is grounded in statistical methods and physics-based modeling, for these key reasons:

1. AI models are black boxes. Their inner workings are often opaque, making it difficult to trace how outputs are derived from inputs —
   a critical limitation for high-stakes energy forecasts that drive financing for multimillion-dollar projects.
2. Physics-based models generalize better and hold up under scrutiny. AI models can overfit to training data or miss edge cases not represented in historical datasets. Training data also tends to be privately held, limiting its utility toward an industry-encompassing training dataset.
3. Finance requires explainability. Uncertainty analyses determine the project’s financial risk and inform its financial structure. In project finance, every assumption must be defensible, not just to the modeler, but to lenders, owners, and independent engineers. It will be nearly impossible to justify why AI made certain choices that led to specific answers.

Plan sets and permitting

AI is deeply integrated into Wattmonk’s products and services. Their internal design teams use AI to accelerate plan set creation, while Zippy, their upcoming plan set automation tool, enables real-time edits, electrical calculations, and instant line diagrams — redefining solar design workflows. The company’s Plan Set QC, AHJ Lookup, and Document Verifier have helped reduce delivery times by up to 4x.

“Our Plan Set QC Agent leverages vision-based AI to ensure designs meet local codes by reading and validating complex diagrams and tables,” says Ankit Sheoran, founder and CEO of Wattmonk. “In our ordering platform, AI automates data extraction from documents, making the process faster and less manual. Site surveys, historically time-consuming, are now completed 30 to 45 minutes faster on average with Framesense, our visual AI model.” They’re still collecting data for Zippy, but Sheoran is seeking an 8x improvement in time taken for plan set creation and revisions.

SiteCapture is building an AI assistant that automates the photo review process by instantly identifying missing documentation and potential quality issues. If there are additional tasks to complete or issues to address, field crews are notified immediately so they can make fixes while they are still on site. The beta version of this feature will be released over the next few months.

“Reviewing field data / photos is critical in any solar project to verify completion of work, perform remote QA / QC, and ensure compliance requirements are met,” says Kamal Shah, SiteCapture founder and CEO. “We expect adding AI to SiteCapture’s jobsite documentation and QC capabilities to reduce the overall cost of the photo review process by at least 50% by eliminating many tedious, repetitive manual steps, reducing human error, and minimizing extra site visits to address issues and collect additional information.”

An American-Made Solar Prize entrant, Eighth Generation Consulting, is developing a solar asset management software using AI, computer vision, geographic information systems, and permitting data to streamline how solar is permitted, serviced, and decommissioned. Through cradle-to-grave tracking, this solution will enable accurate and cost-effective management and decommissioning of solar assets.

Enverus Instant Analyst is a popular tool for large project development. In 2024 alone, Enverus clients were responsible for 44% (23 GW) of all new U.S. solar and wind capacity. This AI-powered tool rapidly surfaces relevant analytics, helping users identify optimal locations, assess infrastructure proximity, and reduce siting timelines, boosting efficiency and confidence in early-stage decisions. By their metrics, Enverus clients have a 9x higher likelihood to reach operation.

“Our generative AI tools, like PRISM Instant Analyst, reduce complex data analysis from days to minutes, enabling faster, smarter decisions,” says Sarp Ozkan, VP of product at Enverus. “With our machine learning models, we now predict project success rates in interconnection queues with up to 90% accuracy, helping developers navigate a landscape where only ~10% of projects typically reach operation.”

Customer service

Bohdi’s AI Assistant is a solar-specific 24/7 virtual customer support tool, helping solar companies stay responsive and efficient even as their customer base grows. It is trained on tens of thousands of real conversations between homeowners and solar companies, documentation from top inverter brands and solar company-specific FAQs. Launched in Q2, AI Assistant has successfully answered over 60% of customer questions and deftly escalated the remainder.

Solar installer Freedom Forever launched Raya this year, a proprietary AI-powered assistant. Raya supports their nationwide sales partners through voice, text, and messaging—offering 24/7 project updates, instant ticket resolution, and information on requirements and timelines. It is built into LIGHTSPEED, their in-house CRM and operations platform.

“In addition, AI is driving document scanning and approval, and helping us automate internal project summaries, reducing the burden on sales and support teams,” says Zachary Bloom, Chief Technology Officer at Freedom Forever. “Unlike generic chatbots, Raya was purpose-built for the solar lifecycle and deeply integrated across Freedom’s tech stack.”

What once required multiple handoffs and manual checks is now resolved instantly through Raya’s connection to real-time workflow data. This improvement has led to a 100% reduction in ticket backlog compared to pre-2022 benchmarks, according to Bloom.

Monitoring and quality control

Omnidian, a dedicated third-party provider of asset performance assurance services in the residential and commercial solar energy space, leverages large historical performance datasets of assets under its management, and AI helps optimize performance monitoring, issue detection and diagnostics, remediation of solar assets, and portfolio-level projections. Example: Using deep learning models to diagnose soiling and recommend when clients should clean their panels.

“Soiling impacts are highly regional and weather-specific, but exhibit distinct patterns in our data,” says Gareth Walker, senior manager, data science and analytics for Omnidian. “We have trained proprietary AI to recognize these patterns and categorize them as soiling events. We use this technology to monitor soiling rates at over 70,000 locations across the United States in near real-time.”

At Sinovoltaics, AI plays a key role in its quality assurance services. Sinovoltaics’ in-house tool, SELMA, uses AI to analyze electroluminescence (EL) images of solar modules during factory inspections. SELMA handles image analysis in just seconds, identifying more than 15 types of cell-level defects with over 99% accuracy. They’ve also developed ways for AI to support real-time diagnostics during BESS factory acceptance testing, which can evaluate any thermal or capacity deviation in every battery pack.

“This allows us to detect micro-cracks and other subtle defects that are difficult to catch with the human eye,” says Arthur Claire, director of technology at Sinovoltaics. “What makes it especially powerful is its consistency and speed. AI can scan thousands of modules far more quickly and accurately than manual methods. That means fewer warranty claims, higher reliability, and ultimately better project outcomes for our clients. It’s not just about efficiency. It’s about raising the bar on what quality control in solar should look like.”

On the insurance side, kWh Analytics is using AI to solve specific problems related to underwriting renewable energy assets. Their capacity management system provides real-time insights into risk aggregation across portfolios to make quick calls on capacity allocations.

“Renewables offer several unique challenges to automation: non-standardized data formats, complex site-specific factors, and rapidly evolving technology require nuanced understanding,” says Jason Kaminsky, CEO of kWh Analytics. “AI’s true value comes from using it as a smart assistant, letting it handle the heavy lifting of data processing while allowing experienced underwriters to focus on decision-making. We use generative AI tools to help us process and analyze unstructured datasets to feed our data science models, and separately to help our underwriters evaluate submissions more thoroughly and quickly.”

Training

American-Made Solar Prize semifinalist PowerTechs is developing a reskilling and skills assessment platform for the renewable energy workforce that combines AI and extended reality. This concept could help address the solar workforce shortage and accelerate deployment of solar energy.

Microgrid control

The quality of forecasts for electricity demand and renewable energy has improved dramatically thanks to AI-enabled forecasting models, says Michael Stadler, cofounder and CTO at Xendee:

“Renewable PV production output forecasts for a full week only show a 4% deviation in comparison to the real observed data. In other words, project outcomes can be predicted to within 4% accuracy, which is a huge benefit to project owners that need the confidence to make decisions based on projected outcomes like ROI. Another example, for residential load forecasting (below 10 kW, with very random load spikes), the forecast precision for a full week is 5.5% on an energy basis.”

Microgrid design and control programs benefit in a big way. Xendee has developed and modified multiple AI libraries that its OPERATE platform uses for different control use cases. “OPERATE calibrates itself automatically based on specific performance criteria and use cases,” Stadler says. “Different building types, different locations, and different use cases require different AI models to achieve the highest performance and outcomes.”

Viridi uses AI to run microgrids and fail-safe energy storage systems. The AI learns the energy consumption patterns and the typical production from on-site renewable systems. It then predicts future consumption and production to better optimize the available stored energy.

“AI is used to continuously optimize Viridi’s many algorithms, which are influenced by constantly changing external factors such as weather, site energy needs (load), real-time pricing, among others,” says Jon M. Williams, CEO and founder of Viridi. “The AI algorithms also overcome the challenges of capacity and charge / discharge rates typically seen in energy storage systems during peak shaving and time-of-use shifting events.”

Paired Power provides solar canopies as well as energy management software that manages the solar, EV charging, ESS, and grid power as a microgrid. AI is crucial to ensure that the user of the energy output has adequate supply. “Our energy management software uses AI and machine learning to enable the system to automatically learn and adapt to be able to deliver optimal energy to loads based on the energy sources that are available (solar, storage, and/or grid),” says CEO Tom McCalmont.

Inverters and energy storage

EcoFlow is integrating AI into its whole-home power solutions through EcoFlow OASIS, a smart home energy management system. OASIS leverages predictive analytics to forecast energy consumption, solar generation, electricity pricing and weather conditions. AI synthesizes this data and adapts to energy usage patterns to offer recommendations and smart automations to enhance energy efficiency or improve preparedness for weather-related outages.

“For example, before severe weather events, OASIS proactively sends storm alerts and automatically charges the batteries to 100% using the most cost-effective energy source, stretching backup power for as long as possible,” says Felipe Burga, North American service manager for EcoFlow.

APsystems is embedding AI into its entire ecosystem, from pre-sales to installation to ongoing operations. Just one example on the product side is their “BESS AI” model, which uses deep learning to analyze solar production, energy usage, and regional electricity pricing. It then generates a custom battery charging strategy for each home, which is automatically sent to the user’s energy storage system. Enphase is also integrating AI with IQ Energy Management. It uses AI-driven analytics to monitor energy production and consumption in real-time. Hoymiles has a new version of its monitoring platform that integrates an AI-powered S-Miles Designer tool. It can help simplify system planning and speed up installation.

Solar trackers

Nevados is pioneering a new digital experience for customers, which includes the ability to manage the initial design and bid process while tracking changes and revisions in real-time, along with their impact on costs. “While not incorporated just yet, our vision is for AI to expedite this process even further, allowing for iterative designs and bids in a fraction of the time it takes today,” says Jenya Meydbray, chief commercial officer at Nevados.

ARRAY utilizes machine learning and AI within its SmarTrack suite of products. SmarTrack models are implemented and trained on site-specific conditions, optimizing tracker row angles for maximum energy production during periods of low sun height, when terrain sloping effects impact row-to-row shading, and during periods of overcast or cloudy conditions for diffuse irradiation capture.