Those of a certain age will remember when the servers supporting an organisation’s IT infrastructure were located on site in their own special room.

The advent of the Cloud, which allows companies to “rent” computing power and storage off-site, fundamentally changed this practice, and companies of all sizes now use the services provided by remote global data centres.

Those signing up to the cloud usually get a healthy credit balance to start them off. But once that’s used up, they start paying, and the bill can run into thousands of euro a month.

It’s a hefty overhead and not one that companies are particularly good at controlling, according to Rakhi Gupta, cofounder with Tahir Rabbani of the agentic AI platform Infralign Smart Solutions, which is developing a tool to tackle cloud waste.

“Enterprises waste enormous sums of money on cloud computing, and most existing tools only point out where the money is leaking and then leave it to the company’s own engineers to do the fixing,” Gupta explains.

“Infralign goes the crucial step farther. Our AI reads the company’s actual cloud bill and architecture and writes the code changes that will fix the overspend at its source. A human engineer then reviews, approves and signs off on every change. In short, others make recommendations; we ship the fix.”

The idea for Infralign was born from Rabbani’s experience as the chief architect in a fast-growing AI energy company.

The company’s cloud costs were high, so he put a programme in place that succeeded in reducing the bill by €750,000 a year. However, even with a dedicated team of engineers on the job, it took eight months to bring the bill down in any significant way.

“Much of the work was unavoidably manual: working through the estate service by service, piecing the analysis together by hand and turning that into concrete fixes,” Gupta says.

“It worked, but the effort it demanded was the real lesson. The hard part of cloud cost control was not only spotting the waste but also the slow, manual grind of fixing it. This is exactly why so many identified savings never get implemented by companies.

“Our proposition for Infralign was simple: if we could automate both the analysis and the execution then work that would take a skilled engineering team months to complete could be done in a matter of weeks.”

Infralign was set up in February of this year and is based at the Nova UCD innovation hub in Belfield. For now, the cofounders are working alone with the help of two interns studying for master’s degrees in computer science.

The company is preparing for a pre-seed funding round of €500,000, and once that is in place, formal hiring will begin. As of now, the cofounders have spent time, not money, developing their system, which has been completely built in-house.

Gupta has a PhD in intellectual property and innovation from the Erasmus University in Rotterdam, and she lectured at the National College of Ireland before moving into deep tech and cofounding Viztronics, a company building 3D vision sensors and software for industrial inspection and robotic automation.

Rabbani’s PhD is in 3D image processing, and his background is in enterprise cloud engineering and technical due diligence for acquisitions.

“Tahir has an unusual pairing of skills, as he is someone who can both run a large cloud estate efficiently but also walk into an unfamiliar one, read it at speed and pinpoint exactly what’s wrong,” Gupta says. “Automating that expertise, diagnosing a company’s cloud and turning the findings into fixes an engineer can ship is precisely what Infralign now does.”

Gupta says the company’s competition comes from existing software tools and consultants providing advice on cloud cost savings. The problem, she says, is that both lean primarily on billing data alone.

“Their recommendations are built from surface-level billing with no real grasp of a company’s architecture or what the business actually needs. So, they’re often wrong, they break things when you apply them or they’re quietly ignored,” she says.

“We ground every recommendation in a company’s own bill, its own architecture and its own business context. So, it’s the right change, not a generic one, which we then deliver as an actual, engineer-trusted code change, with a human in the loop every time.”

The big names in cloud provision are Microsoft (Azure), Google Cloud and Amazon Web Services, and their customers are potentially Infralign’s customers too, along with users of other cloud providers. But for now, Infralign is focusing on those running with Microsoft Azure.

“We are going deep on Azure initially rather than spreading ourselves thinly across every cloud,” Gupta says. “Our main customers at this point are Azure-first companies, typically with 100 to 1,000 employees, that don’t have a dedicated in-house cloud-cost (FinOps) function. The buyers of our service are usually CTOs, heads of engineering or platform, and CFOs.”

Infralign has signed its first paying customer and is interested in hearing from companies willing to become pilot sites for its unfolding technology.

Big organisations clearly have a lot to gain from cutting their cloud bills but Gupta stresses that the system works for companies of all sizes and can be of particular relevance for those that don’t have in-house engineering teams and for high-growth, mid-sized businesses looking to streamline cloud cost management.

Infralign will make its money in two ways: through a subscription based on size and a gain-share model whereby it will take a percentage of any savings made using its system.

“If a company does not save, then we don’t make any money,” Gupta says. “We are starting with the Irish and UK markets before expanding across northern Europe, but we have global ambitions. Cloud overspend is a worldwide problem, and the product works anywhere Azure does,” says Gupta, whose company is participating in the AI ecosystem accelerator programme run jointly by Nova UCD and CeADAR, Ireland’s centre for applied AI.