The Center for Data Innovation recently spoke with Nicolas Gutierrez, Chief Operating Officer of Invertix, a San Francisco-based company developing AI agents for renewable energy operations. Gutierrez explained how Invertix helps operators monitor energy assets, identify equipment problems, and automate routine tasks so teams can manage growing portfolios more efficiently.
David Kertai: What challenges in energy operations is Invertix solving?
Nicolas Gutierrez: Renewable energy portfolios, collections of energy facilities and equipment managed by an operator, continue to grow faster than operations teams, forcing operators to spend much of their day moving between monitoring systems, spreadsheets, emails, and reports to understand why a facility is producing less energy than expected. Portfolios also often include equipment from different manufacturers, multiple monitoring platforms, inconsistent data formats, and gaps in available information. These challenges make it difficult to maintain a complete picture of how each facility is performing. As a result, teams struggle to identify problems quickly, prioritize maintenance, and manage larger portfolios without adding staff.
Invertix addresses these challenges with a platform that uses AI agents to automate renewable energy operations. Connected to existing monitoring and operational systems, the platform helps teams monitor assets, analyze performance data, identify the causes of energy losses, prioritize issues, and assist with maintenance workflows. By automating repetitive analysis and investigation tasks, Invertix helps operators respond to problems faster and focus on decisions that require human expertise.
Kertai: How does Invertix integrate with existing software and monitoring systems?
Gutierrez: Most renewable energy operators already rely on monitoring systems, so replacing them is rarely practical. Instead, we connect to the tools they already use. Through read-only access, we gather data from supervisory control and data acquisition systems, monitoring portals, internal databases, and equipment manufacturers’ platforms. We then bring that information together into a single operational view.
The platform only retrieves data and does not modify operational systems, which helps information technology and cybersecurity teams review and approve deployments more quickly. As a result, customers can typically begin using the platform in weeks rather than months.
Kertai: How do your AI agents identify equipment problems and recommend solutions?
Gutierrez: The process begins by comparing actual energy production against expected production based on weather conditions and site characteristics. When the system identifies a gap, the AI agents investigate, much like an experienced engineer would. They review equipment alerts, compare performance across nearby devices, examine weather conditions, analyze historical trends, and review previous maintenance records. They then identify the most likely cause and recommend an appropriate next step.
For example, the system might recommend continued monitoring, cleaning solar panels, remotely resetting equipment through existing procedures, or dispatching a technician. It also estimates the financial impact of each issue, helping operations teams prioritize problems that cause the greatest energy losses.
Kertai: Which operational tasks can AI agents handle independently, and which still require human oversight?
Gutierrez: Our principle is straightforward: AI agents perform the repetitive analysis, while people remain responsible for decisions that involve operational risk or financial consequences. The agents continuously monitor assets, organize information, investigate anomalies, and prepare recommendations. This allows operators to spend less time gathering data and more time evaluating potential actions.
Human oversight remains essential when a decision could affect equipment performance, maintenance spending, or regulatory obligations. Actions such as dispatching technicians, escalating issues to equipment manufacturers, or authorizing operational changes stay under human control. Each customer sets those boundaries based on their own risk tolerance and procedures. As teams gain confidence in the system, they can choose to automate additional workflows while retaining final authority over critical operational decisions.
Kertai: What steps do you take to ensure your AI agents make accurate recommendations?
Gutierrez: Transparency is a critical part of building trust. Every recommendation includes supporting evidence, such as the underlying data, performance comparisons, and the reasoning behind the conclusion. This allows operators to quickly understand why the system recommended a particular action rather than relying on a black box.
Before deployment, we validate recommendations against each customer’s historical operational data and existing processes. This allows us to evaluate performance against real operational outcomes before the platform becomes part of daily workflows. We also maintain a continuous feedback loop. Operators can correct recommendations, and those corrections help improve future performance. By tracking which recommendations prove accurate and which require adjustment, we continuously refine the system while helping operators build confidence in its recommendations.


