Our verdict
Fusemachines is an AI development and consulting company in New York, founded in 2013 by Sameer Maskey, a Columbia University computer science PhD who has also taught there as an adjunct professor. It sells its own AI platform, AI Studio, along with ready-made AI tools and agents, but its annual report says product sales are an insignificant share of revenue. Most of its income comes from engineers who customize, build and run AI systems inside client companies. Much of that work depends on its operations in Nepal, where Maskey started an AI Fellowship program in Kathmandu to train engineers. Fusemachines has traded on Nasdaq since October 2025, reported $7.71 million in revenue for 2025 and had more than 270 full-time employees at the end of that year (Form 10-K; About; Columbia Computer Science).
Its most detailed recent project is a predictive maintenance system for a US natural gas compression operator. Fusemachines moved the client's sensor data from Databricks and Dataiku into Microsoft Fabric, trained a model on engine anomalies that the client's equipment specialists had checked, and now flags developing engine faults hours in advance, with an AI-written summary of likely causes for each alert. The company says the system covers more than 1,000 engines. For a global retailer it built demand forecasting models that it says cut forecast error by up to 50% against the retailer's older statistical methods and lowered the estimated chance of store stockouts by about 25%, and it automated price elasticity estimates for thousands of products. It also runs an email screening tool for a commercial real estate bridge lender that sorts about 500 deal submissions a month (compressor case study; forecasting; price elasticity; bridge lender). None of these case studies identifies the client.
Fusemachines suits a retailer, consumer brand or industrial operator that wants forecasting or predictive maintenance built on its own data by a team with offshore costs and US management. Buyers planning a multi-year engagement should weigh its finances: its auditor raised substantial doubt about its ability to continue as a going concern, and revenue fell 12% in 2025.
The right buyer
- Industrial and energy operators that want early warnings from equipment sensor data, like the natural gas compression operator whose fault detection system covers more than 1,000 engines on Microsoft Fabric.
- Retailers and consumer brands that want demand forecasts and price elasticity estimates fed into the dashboards their planners already use, as in its forecasting and pricing work for a global retailer.
- Companies that want lower-cost AI engineering with US leadership; most of its delivery runs from Nepal, where it has trained engineers through its AI Fellowship since 2017 (Columbia Computer Science).
- Commercial real estate lenders that want inbound deal emails sorted against their lending rules, using the LenderIQ engine and Outlook inbox tool it built for a bridge lender.
Reasons to look elsewhere
- Its auditor's report on 2025 results raised substantial doubt about whether Fusemachines can continue as a going concern; it ended the year with about $4.2 million in cash after using $5.5 million in operations (Form 10-K).
- Revenue fell 12% to $7.71 million in 2025 after 12 customers ended their contracts and 16 reduced them, and two customers made up about 30% of revenue (Form 10-K).
- One of the client testimonials on its homepage comes from Bharat Krish, who joined the Fusemachines board when the company went public (press release).
- Its case studies do not identify their clients, so buyers cannot check the results with the companies involved (example).
Scores
| Criterion | Weight | Score | Basis |
|---|---|---|---|
| Production track record | 20% | 7.2 | Predictive maintenance across more than 1,000 compressor engines; demand forecasting and price elasticity work for a global retailer over several engagements; deal email screening for a bridge lender. |
| Technical depth | 15% | 7.3 | Supervised fault detection with AI-written alert summaries on Microsoft Fabric; deep learning forecasting; earlier speech recognition and entity extraction for medical records; PhD-led founding team. |
| Industry depth | 10% | 7.0 | Repeat work in retail and consumer goods, energy equipment and commercial real estate lending; earlier healthcare and government projects. |
| Delivery model and handoff | 15% | 6.9 | Forward deployed engineers who work in the client's systems and stay on after launch; delivery depends heavily on its team in Nepal. |
| Commercial clarity | 10% | 6.5 | No rates or package prices published; offers a fixed-scope AI and data assessment before larger work; clients can buy software, services or both. |
| Platform independence | 10% | 6.8 | Builds on Microsoft Fabric, Power BI and Azure for clients and promotes open-source language models, but also sells its own AI Studio platform. |
| Stability and scale | 10% | 5.4 | Operating since 2013 with 270+ employees, but 2025 revenue fell 12% to $7.71 million, two customers made up about 30% of revenue and its auditor raised going concern doubt. |
| Independent evidence | 10% | 6.0 | Audited SEC filings give unusual financial transparency; client praise comes from its own site and none of its case studies identifies the client. |
| Overall | 100% | 6.7 | Weighted average |
Where Fusemachines fits
| Need | Fit | Why |
|---|---|---|
| Predictive maintenance from equipment sensor data | Strong | Fault detection across more than 1,000 natural gas compressor engines, with alerts reviewed by the client's specialists. |
| Retail demand forecasting and pricing analytics | Strong | Forecasts and price elasticity estimates for thousands of products feeding a global retailer's planning dashboards. |
| Generative AI screening of email or documents | Good | Outlook inbox tool that sorts and summarizes about 500 lending deal submissions a month. |
| Buyer who needs a financially secure long-term vendor | Weak | Auditor raised substantial doubt about its ability to continue as a going concern in its 2025 annual report. |
What clients and leaders say
“Fusemachines revolutionized our regulatory compliance product with a data pipeline that efficiently handles healthcare price information.”
“Fusemachines has been a pivotal technology partner for us for many years in various areas of technology stack.”
“Historically, we have incurred recurring net losses from operations and negative cash flows from operating activities.”
Who leads Fusemachines
Sources (10)
- Fusemachines: homepage and client testimonials
- Fusemachines: About us and leadership
- Fusemachines: Form 10-K for 2025 (SEC)
- Fusemachines press release: closing of business combination and Nasdaq listing
- Fusemachines: Compressor engine fault detection case study
- Fusemachines: Demand forecasting for a global retailer case study
- Fusemachines: Price elasticity case study
- Fusemachines: High-volume bridge lender case study
- Fusemachines: Medical records speech recognition case study
- Columbia University Department of Computer Science: Fusemachines and its AI fellowships (2017)