Autonomous systems have impacted how we design, build and maintain the world’s largest critical infrastructure assets. AI Clearing has emerged as one of the clearest examples of that change by turning drone imagery, design data and field reports into real time project intelligence for solar, transmission, pipelines, roads, rail…and more.
From Asset Heavy Problems To AI Powered Answers
When California’s wildfires exposed a critical vulnerability for utilities facing thousands of miles of damaged lines and poles in 2014, Michael Mazur, AI Clearing CEO, identified a need to fill with emerging tech. The slow and human intensive conventional approach required teams to walk corridors, climb structures and manually document conditions. This consumed months of time, under extreme pressure. “Nearly all my clients had an army of people whose only role was to walk, look and check the infrastructure,” Mazur said.
At the same time, a workforce shortage loomed. With far more engineers retiring than universities could replace, a capacity gap widened. Mazur realized drones could capture the scale of the problem while computer vision could compress the analysis, without the heavy manpower lift. “I came up with an idea to fly drones over thousands of miles of assets, take pictures and then use AI to analyze it,” he recalled. “Instead of spending six months or so with a large team of people looking at those pictures, we could do it in two weeks time.”
Before founding AI Clearing, Mazur had spent his career in technology consulting, bringing emerging tech into asset heavy industries such as utilities, gas and oil and transport infrastructure across more than 40 countries. That experience that gave him a front row seat to the limitations of traditional inspection models and the demographic crunch in engineering.
In January 2020, Mazur and co-founder and CTO Adam Wiśniewski launched AI Clearing and designed an AI-native platform from scratch, specifically for complex infrastructure projects. Through the disruptions of the pandemic, they focused on building the foundations of a deep, clean data lake and a product architecture that would make sense to project managers looking at hundreds of miles of work.
Today, the combination of autonomous sensing, combined with AI Clearing’s intelligent processing, has redefined inspection for asset heavy industries, from utilities to oil and gas to transport infrastructure. The company now operates out of both Austin, Texas and Warsaw, Poland. Over the years, it has racked up an impressive list of global energy-sector and infrastructure clients.
Building The Data Lake That Teaches AI To See Infrastructure
Construction imagery at scale simply did not exist in labeled form. The team paid out of pocket to collect drone data from solar sites across Canada, the United States, the Middle East and Europe, then expanded to wind, roads, railways, pipelines and transmission lines.
“The first and most important thing to succeed is the data lake,” Mazur explained. “A collection of different pictures of assets, materials, machinery properly annotated, so that every time there is a picture of a solar panel, it’s annotated as solar panel, whenever it’s a wire, it’s a wire, whenever it’s a trench, it’s a trench.”
AI Clearing’s Warsaw R&D center now houses roughly fifty AI scientists, developers and engineers dedicated to building and maintaining that lake. Their labeling guide for solar alone runs 300 to 400 pages, Mazur said. It specifies exactly how to mark each material and activity in every phase of construction.
The scale of this data lake is difficult to visualize. Mazur offered a physical analogy. “Imagine a mile wide strip of land stretching from San Francisco to New York, back to San Francisco and halfway back again,” he explained. “That is the current size of AI Clearing’s annotated coverage.” It includes more than 6000 drone inspections and many millions of labeled objects. The team has re-annotated the entire corpus multiple times as they learn what clients need.
Building this AI was not easy. Edge cases caused challenges, from which the team learned. Snowfall on a solar farm near Calgary changed the familiar green landscape to white, which cut model accuracy from above ninety percent to fifty until the team retrained for snow backgrounds. Intense sun in Spain’s Extremadura region caused glare on panels that confused the model until reflections became part of training. In Saudi Arabia, a subcontractor stored single panels under steel racks before installation, making them look to the AI like fully mounted modules. “We had to add height as a learning channel,” Mazur noted. “So now the shape and height of things are part of the model.”
Company co-founder Adam Wiśniewski estimates that roughly 250,000 engineering hours have gone into building their system, from data quality checks and 3D model generation to report pipelines and dashboard logic. The result is an AI that does not just recognize objects but also understands the activities and constraints that define infrastructure projects.
CORE: A Browser Window Into Real Projects
AI Clearing’s CORE platform operates as an intelligent operating system (OS) for large scale infrastructure and renewable energy projects. It runs entirely in the browser. When a superintendent or project manager logs in, they see a detailed, drone-derived 3-D map of their site on one side of the screen and a rich panel of AI-generated metrics and dashboards on the other.
As drones fly weekly or twice weekly missions over solar farms stretching two miles, transmission corridors running fifty miles, gas pipelines and highway segments spanning dozens of miles, CORE ingests those images. It then constructs a 3D model of each site and assigns every pixel to a specific class, such as solar panel, pile, trench, cable, concrete, pavement, fencing and more.
Once the system identifies what is on the ground and where, it starts counting and comparing. It calculates how many panels were installed since the last flight, how many miles of pipeline sit in trench, how much roadbed is complete and how those measurements line up against the design imported from BIM and CAD tools like Autodesk or Bentley. “The system compares what’s built consistent with the design and flags any discrepancies,” Mazur says. Those early misalignments are often the seeds of future budget overruns. Assets built in the wrong place or in the wrong way frequently need demolition and rework.
CORE also integrates schedules from project planning systems such as Oracle Primavera. It checks which tasks are physically completed against their planned timelines and highlights slippage before it becomes a crisis. For a typical infrastructure project, the path from drone capture to AI Clearing dashboard takes about five hours. This replaces a week or more of manual review with near real time visibility. That matters when owners and contractors have fewer than ten supervisors on site and can realistically see only five to ten percent of the work each month by walking it.
“Up to now, the key method of controlling those projects was a prayer,” Mazur quipped. “But ninety percent of projects are delayed and over budget.” CORE gives those teams a way to move from guesswork to evidence and measurable progress.
Clara: Conversational AI For Linear Projects
As infrastructure projects and datasets grow more complex, AI Clearing has pushed beyond dashboards into conversational interaction. Clara, the company’s agentic interface, currently operates as a voice and text assistant wired directly into CORE. “Clara is a voice interface which allows you to conversate with data,” Mazur elaborated. A project manager can ask, “Hey Clara, on model 75 of this highway construction, tell me if there is an issue and what was the progress,” and Clara will review the relevant dashboards and respond.
This provides huge dividends, particularly for linear infrastructure like transmission lines, pipelines and roads, where site teams spend hours driving trucks through remote areas. Checking a phone is unsafe and often impractical. Hands free conversation with a project’s source of truth becomes both a safety feature and an operational advantage.
Mazur and his team plan to take Clara into project documentation next. Large infrastructure programs generate thousands of PDFs in the form of contracts, addenda, technical specifications, warranties and regulatory notices. Not everyone on a project reads every line. Clara will increasingly cross-reference those documents against what is happening on the ground. In one scenario Mazur described, Clara may say, “On the project next to Seattle, there was a contractor installing panels yesterday and it was forty degrees. According to the specs, you cannot install them if it’s below forty five degrees. Should I draft an email to Legal?”
On the warranty side, the workflow gets even more direct. A supervisor standing next to a broken panel takes a photo and asks Clara to prepare a warranty claim. Because CORE already knows the exact geolocation and identity of that panel, Clara can connect the image to the right serial number, warranty terms and supplier, then automate notifications and documentation that would normally take hours at a desk. “Here, it’s just done,” Mazur said. “And you can finish earlier.”
Compliance And Claims In A Single Digital Record
As powerful as automation can be, infrastructure remains a domain with strict safety, regulatory and contractual requirements. AI Clearing’s platform supports those obligations. Drone data and AI models handle progress tracking, geometric checks and quantity verification, but many safety and commissioning tests still must be performed in person and documented according to law.
“Our platform allows and supports this process,” Mazur said. Field inspectors use AI Clearing’s tablet apps to view the full site, see their location over the drone derived model via blue dots and open asset specific test forms while standing next to a component. They conduct required checks, sign digitally and push those records back into the system, which stores a unified, traceable record of inspections and approvals.
Bringing AI and human workflows together pays dividends when something goes wrong. When tornadoes hit Texas last year, three AI Clearing clients used the platform’s records to file and win full insurance claims quickly, because they could provide evidence of conditions and quantities before the event. The same objectivity helps owners and contractors avoid costly issues in subcontractor reporting. Sometimes firms “are too optimistic or count things twice,” Mazur noted. The platform’s measurements ensure invoices match reality.
For Mazur, though, the most important benefit still lands on the people in the field. “You are freeing up the time of your superintendent,” he explained. Instead of spending hours checking whether 1000 panels are installed, supervisors focus on the roughly 400 flagged with issues. Crews report saving up to two hours a day as a result. “These people are not working eight hours. They’re working long days, separated from their families. Freeing up two hours is an amazing thing.”
Looking Ahead: Autonomy For The Entire Project Ecosystem
AI Clearing sits at the intersection of autonomous systems, AI and critical infrastructure at a moment when demand for new assets far outpaces available engineering talent. By pairing drones with CORE and Clara, the company has extended autonomy in the field from individual vehicles to the entire project ecosystem of designs, schedules, contracts and inspections.
Mazur and his team want to help builders and asset owners deliver more infrastructure, with higher assurance and fewer surprises, using teams that are already stretched thin. As AI-native platforms like AI Clearing become standard practice across solar, transmission, pipelines, roads and rail, the industry’s most critical assets will be monitored continuously by machines, with humans focused on the decisions that matter most.