Kashmir Intelligence and the future of AI in Energy

Six years ago I was asked by the 英国帝国理工学院 Enterprise Labs team to be one of their Experts in Residence, helping students commercialise research projects, and I jumped at the chance, there is nothing that gets me more excited than building technology companies with smart people, there’s also a deep truth in the maxim, “If you can’t teach it, you’ve not learned it”, so a reason to constantly drill fundamentals was welcome.

In one of my first sessions as an expert I met the most gifted technologist and thinker I’ve met in my twenty years in this industry, Samyakh (Sam) Tukra . He was taking-to-market a technology product that predicted mortality probability in ICU patients allowing Doctors to immediately help those in most need without wasting vital hours on lab work and chart interpretation. We took his product into 8 NHS hospitals where it worked fantastically, but there was a huge problem, there was no commercial path for his technology, no buyer, no tangible value to be demonstrated for the hospitals, and no existing solution to replace, a perfect storm of market variables destined to ensure a product goes nowhere. This was a hard pill for both of us to swallow, but there was no choice, Sam had completed his PhD and had to start earning rent money. He made one of the hardest decisions and entrepreneur has to and divested his technology (it’s still in use today) and went out into full-time employment determined to, and I quote Sam here, “find the biggest problem on earth”.

He found it.

“They need much, much better AI” ... that’s how our conversation began 3 years ago.

Sam was calling me from a downstream facility in Europe, he’d been building computer vision products for energy companies for a couple of years and had recently sold a patent to the company that he was now contracting for, what he’d found turned out to be the biggest and most valuable problem that exits today.

There is no future of our species without a transformed energy industry, and I’m not just talking about growing the use of wind, solar and hydrogen etc. these energy sources will help move us towards sustainability, but if we want an abundant, sustainable future with affordable energy for all, we have to deliver advanced technologies into downstream environments.

??Global energy demand will grow by over 50% in the next decade

??We will add 2.2B people to the planet in the same time period

??Global energy supply over this same time period will only grow by around 30%

>70% of our planet’s energy, agricultural, manufacturing and transportation reliance is provided for by downstream industry. <

If we're to meet this energy production shortfall we need to bring advanced technology into downstream environments.

An industry that is being hindered by a particular technological approach.

There are immutable truths when it comes to high performing deep learning in industrial environments.

1. Data volume

2. Data quality

3. Data recency

If you work with all of the data being collected, in high quality representations, with the most recency possible, you will deliver the highest quality inference attainable.

Current cloud and centralised approaches can't do this. They’re too expensive (data transfer on 100% of available data from a Lvl 6 location would cost around ￡25M to transfer to the cloud), too slow (months to build a successful model), too risky from a data security perspective (this is the most valuable data on earth), and too talent intensive to deliver the performance required to drive new benchmarks of production margin and efficiency in heavy industry.

We need a new type of infrastructure layer built for heavy industrial environments instead of trying to force performance out of cloud-based approaches.

This is why three years ago Sam and I began work on Kashmir Intelligence , and what led us to develop Orbital, our patented edge infrastructure layer that delivers cloud-beating deep learning capabilities into downstream environments at a fraction of the cost, with none of the security risks.

We’re excited to be emerging from stealth with real world performance underway at Exxon/Gruppo api and ABB/ICL facilities and are excited to formally launch Orbital at Global Refining & Petrochemicals Congress 2025 in Delhi on the 27th of June as part of our Indian market focus. We’re also fortunate to be backed by exceptional venture funds in Stride.VC , Jigsaw.vc , Tiny, Material, Firedrop and Truesight as part of our Pre-Seed round of $5M, as well as supported by an incredible and committed team of advisors in Bill Kelleher , Anil Bharath , Greg Gabel , Samarth Derdekar , and Dr Divakar Kamath .

To learn more visit us a kashmirintelligence.com and stay tuned for more information and company news.

Big Data needs a sharp Edge, and we’re looking forward to showing you what Orbital can do.

kashmirintelligence.com