Karen Nikoghosyan runs three companies from Gyumri, Armenia, a city that is not a tech hub by any standard measure. eSwap Global, his main product, is a multichannel commerce platform that connects Amazon, Shopify, Walmart, eBay, Etsy, and Wayfair into a single dashboard. In late 2024, he added an AI suite to the platform: demand forecasting, dynamic pricing, and predictive restocking. The long-term goal is agent-based AI as the operating layer for multichannel retail, where sellers supervise rather than manually operate their businesses.
From CTO to Founder to Platform Builder
Nikoghosyan's career started early. By 23, he was CTO of GTech Solutions, overseeing software delivery for international clients. The role exposed him to a specific problem: companies selling on multiple marketplaces were stitching together fragmented systems for orders, inventory, shipping, and reporting. Each marketplace had its own API, its own data format, its own operational quirks. Sellers were spending more time reconciling systems than selling products.
In 2016, he founded Ghost Services, a software development agency in Gyumri, to build end-to-end products for clients worldwide. The agency has delivered work across e-commerce, healthcare, logistics, and biopharma. It ranked 16th on the Clutch 100 Fastest-Growing Companies list in 2024 and has served more than 80 clients. The agency model gave him revenue and flexibility, but the underlying problem he saw as a CTO was still unsolved: multichannel commerce needed a unified layer.
In 2019, he launched eSwap, initially as an inventory management tool for multichannel sellers. The platform connected marketplaces and web stores through a single dashboard, giving sellers one view of sales, profits, returns, orders, and stock. Over time, eSwap expanded into order management, shipping, accounting, invoicing, and analytics. The integrations grew to include Amazon, eBay, Walmart, Shopify, WooCommerce, BigCommerce, Etsy, and Wayfair.
The AI Operating Layer
The AI suite that launched in late 2024 is the most significant product evolution. Three core capabilities define it: demand forecasting, dynamic pricing, and predictive restocking. The forecasting system uses sales history, seasonal trends, and marketplace activity to recommend reorders and warehouse allocation. The goal is to help sellers avoid both excess stock and stockouts without manual analysis.
Dynamic pricing adjusts prices based on marketplace conditions, competitor behavior, and inventory levels. Predictive restocking uses the same data signals to trigger replenishment before a seller runs out of popular items. Together, these three capabilities form what Nikoghosyan calls an intelligent command center, where inventory, pricing, and fulfillment decisions increasingly happen autonomously.
The practical impact is a shift in how sellers operate. Instead of manually checking inventory levels, comparing prices across marketplaces, and placing reorders based on gut feeling, sellers set parameters and let the AI handle the execution. The human moves from operator to supervisor. Nikoghosyan frames this bluntly: if you are using AI for operations, you can replace your operations department with one AI tool.
Building From Gyumri
The geography matters. Gyumri is Armenia's second-largest city, but it is not a tech hub in the way that Yerevan, or Silicon Valley, or Bangalore are. There is no local venture capital ecosystem, no concentration of talent agencies, no proximity to major markets. Clients are won through results, not proximity or networks.
That constraint forced a specific operating philosophy. Ghost Services had to deliver measurable outcomes because there was no network效应 to fall back on. The engineering team had to be self-sufficient because the nearest major city with a deep talent pool was hours away. The product had to work globally because the local market was too small to sustain a SaaS business.
Nikoghosyan now splits his time between the US business operation and the Armenian engineering team. He leads eSwap's product strategy, AI roadmap, business development, and commercial operations while continuing to run Ghost Services and the academy. The structure is intentional: the engineering stays in Armenia where costs are lower and the team is built, while the business development happens in the US where the customers are.
The Talent Pipeline Problem
While scaling Ghost Services, Nikoghosyan hit the same wall every company in a non-hub city hits: there were not enough trained developers. Rather than relocate or rely entirely on remote hiring, he founded the Techno-Educational Academy in 2017 to create a local talent pipeline. The academy has graduated more than 4,000 students. Alumni have gone on to found more than 15 IT companies in Gyumri.
The stat he cites with the most pride is not the revenue or the client count. It is the 15 companies founded by academy graduates. That is a multiplier effect: one educational institution creating the conditions for an entire local ecosystem. The companies are small, mostly agencies and product shops, but they employ people, build skills, and keep talent in the city rather than losing it to emigration.
The academy also serves a practical business purpose. Ghost Services has a local hiring pipeline that does not depend on competing with Yerevan or remote employers for talent. The graduates are trained in the specific stack and workflows that Ghost Services uses. The feedback loop between the agency and the academy keeps the curriculum aligned with market demand.
Where This Is Heading
The next phase for eSwap is deeper AI capabilities and expanded US presence. The AI suite is functional but early. Demand forecasting works for established products with sales history, but new products with no data still require manual planning. Dynamic pricing works within a single marketplace, but cross-marketplace pricing optimization is harder because each platform has different fee structures, different buyer behaviors, and different algorithmic preferences.
The agent-based vision is the endpoint Nikoghosyan is working toward. In that model, the AI does not just forecast or recommend. It executes. It places reorders, adjusts prices, shifts inventory between warehouses, and handles customer inquiries about shipping status. The seller's role reduces to setting strategy and reviewing exceptions. The AI handles the operational complexity that currently requires a team of people.
The constraint is trust. Sellers need to trust the AI with their money, their inventory, and their customer relationships. The trust builds gradually, through a history of correct decisions and transparent reasoning. eSwap's approach, building the AI on top of a platform that already manages the operational data, gives the AI access to the signals it needs. The sellers already trust the platform with their data. The extension from data management to autonomous decision-making is a smaller leap than starting from scratch.
Nikoghosyan's operating principle is consistent across all three ventures: build systems that outlast his involvement. The academy produces graduates who build their own companies. The agency delivers products that clients run independently. The platform aims to run commerce operations with minimal human intervention. The pattern is the same: create something that works without you, then move on to the next problem.