The global supply chain for heavy equipment parts runs on a diesel engine’s worth of friction. A mechanic in a remote quarry needs a replacement turbocharger for a 2017 Caterpillar 797F. The part number is worn off. The local dealer’s catalog is a PDF labyrinth. The mechanic’s afternoon vanishes into a phone tree, while the $5 million truck sits idle. This is the daily grind Heavi AI, a new startup out of Istanbul, says it can lubricate with a purpose-built AI [Heavi AI, retrieved 2024].
Founded in 2025 by solo founder Mert Çelenk, the company is building custom software for heavy industry, with an initial focus on identifying and procuring diesel engine and heavy equipment parts [CB Insights, retrieved 2024]. The claim is straightforward: its AI, trained on schematics and parts catalogs and backed by vehicle identification number (VIN) level intelligence, can help users find the right part 75% faster and at a better price, while coordinating delivery [Heavi AI, retrieved 2024]. It’s a narrow wedge into a massive, stubbornly analog industry.
The bet on diesel intelligence
Heavi AI’s proposition hinges on specificity. A general-purpose large language model can describe a turbocharger, but it likely can’t cross-reference a模糊 serial number against a proprietary OEM catalog to find the exact match for a specific engine block in a specific mining truck. The company says its AI understands diesel engines and heavy equipment, implying a trained model or a structured database that maps parts to machines [Heavi AI, retrieved 2024]. This isn’t a search engine wrapper; it’s a tool built for a trade. The value isn’t in the AI itself, but in the accuracy of its output,getting the right part, the first time, to the right place. For an industry where downtime is measured in thousands of dollars per hour, even shaving an hour off the procurement process can justify a software fee.
An early-stage puzzle
What’s visible of Heavi AI is a clean marketing page and little else. There are no public customer case studies, no disclosed funding rounds, and no named investors [PitchBook, retrieved 2024]. Founder Mert Çelenk, who serves as Co-Founder and Technology Lead, holds certifications in AI for Medical Diagnosis and database systems, and studied at Yıldız Teknik Üniversitesi [The Org, retrieved 2026] [LinkedIn, retrieved 2026]. The company appears to be in its earliest stages, likely pre-seed or bootstrapped. This creates a classic chicken-and-egg scenario for a B2B tool: heavy industry buyers are notoriously conservative, preferring proven solutions from established vendors. To prove its 75% speed claim, Heavi AI needs real-world deployments. To get deployments, it needs to prove its claim.
A secondary, more mundane risk is brand confusion. A much larger, well-funded analytics platform called HEAVY.AI (formerly OmniSci) operates in a completely different sector [Perplexity Sonar Pro Brief]. For a small startup relying on organic search, sharing a name with an incumbent is an unnecessary headwind.
The unit economics of downtime
Let’s run a simple, conservative scenario. Assume a mid-sized mining operation has one large haul truck down. Industry estimates peg the cost of idle mining equipment at anywhere from $100 to $500 per hour. If Heavi AI’s software saves just two hours in the parts identification and ordering process for that single incident, it has created $200 to $1000 of value. If a software subscription costs a few thousand dollars per year, the payback could come from a handful of averted delays. The math gets compelling quickly at scale, which is the bet: that the software’s cost will be invisible next to the cost of the downtime it prevents.
For Heavi AI to matter, it must eventually beat the incumbent,not a software giant, but the current process itself. That process is a combination of dog-eared paper catalogs, a dealer’s institutional knowledge (which retires with the employee), and long-distance phone calls. It’s slow, but it’s known. The company’s task is to prove that its AI is not just faster, but more reliable. If it can do that, it won’t just be selling software; it will be selling back time.
Sources
- [Heavi AI, retrieved 2024] Heavi AI - AI Tools for Heavy Duty | https://heaviai.com/
- [CB Insights, retrieved 2024] Hevi AI CEO, Founder, Key Executive Team | https://www.cbinsights.com/company/hevi-ai/people
- [PitchBook, retrieved 2024] Heavi AI 2026 Company Profile | https://pitchbook.com/profiles/company/1360479-25
- [The Org, retrieved 2026] Mert Çelenk - Co-Founder & Technology Lead at Hevi AI | https://theorg.com/org/hevi-ai/org-chart/mert-celenk
- [LinkedIn, retrieved 2026] Mert Çelenk - Co-Founder & Technology Lead - Hevi AI | https://www.linkedin.com/in/mert-celenk/
- [Perplexity Sonar Pro Brief] Distinction between Heavi and HEAVY.AI | Analysis based on public web search