.webp)
Invertix has joined Y Combinator's Fall 2026 programme and entered the UK market, expanding its work on AI agents for critical infrastructure.
The San Francisco-based business has also opened in the US and Spain as it broadens its reach beyond its existing energy customer base. It works with energy owners and service companies that collectively manage more than 50 GW of energy assets, while noting that this figure refers to customer-managed assets rather than capacity controlled by its software.
Founded by Joseph Perrotta and Kaan Durmaz, Invertix is developing software that combines a data-normalisation platform with AI agents tailored to operational workflows inside infrastructure businesses. Its current work in energy includes monitoring plants, investigating equipment issues and coordinating the tasks needed to address them.
It is now considering whether that model can be used more widely in sectors including data centres, where power, cooling and maintenance coordination are central to day-to-day operations.
Operational focus
Its approach grew out of conversations with 500 people working across critical infrastructure. Those discussions led the company away from an initial concept of plug-and-play agents and towards systems built around each customer's operational context.
Rather than relying on an alert or isolated data point, the platform is designed to connect equipment records, technical documents, maintenance history and internal responsibilities to a specific workflow. Company engineers work with customer teams to map how decisions are made and who is responsible for the next step before configuring agents for particular tasks.
One example is energy maintenance. The company argues that an equipment alarm does not by itself provide the asset's history, the relevant procedure or the person accountable for responding. Those links are needed if software is to support investigations and prepare follow-up work.
For data-centre operators, the proposed application would follow a similar pattern. Possible uses include bringing together equipment data, technical documentation, records of previous interventions and team responsibilities during a facilities investigation. However, the company is not announcing a current deployment with a data-centre customer.
Expansion plans
Invertix originated in Munich within TUM.ai, a student AI initiative where several team members gained research experience. It has since moved to San Francisco for the Y Combinator programme and plans to return to Europe afterwards.
Its expansion is also being supported by Spanish Chief Operating Officer and founding team member Nicolas Gutierrez de Esteban. Investors include Vireo Ventures and Italian Founders Fund, and the company previously announced €1.7 million in pre-seed financing.
Invertix describes its longer-term goal as "Energy Superintelligence", a term it uses for coordinated agents that help organisations operate and expand energy infrastructure across multiple assets and departments. The concept reflects the founders' view that energy systems and AI development are becoming more closely linked, with better-run infrastructure supporting computing growth and AI tools helping companies manage physical operations.
Perrotta said the company's central lesson has been that useful AI in infrastructure depends as much on reliable organisational context as on model performance. "Our initial idea was to offer plug-and-play agents. Those conversations exposed a more fundamental requirement: an agent needs reliable operational context before it can carry out useful work," said Joseph Perrotta, Co-Founder and Chief Executive Officer of Invertix.
That requirement has shaped how the company deploys its system inside customer organisations, with embedded engineers helping define permissions, review steps and responsibilities around each workflow. "Our ambition is for coordinated agents to support and execute work across departments while remaining grounded in how the business operates," Perrotta said.