Wirestock provides an advanced multimodal visual training data platform for AI research and development teams. The company produces rights-cleared datasets across the model training pipeline while connecting photographers, videographers, designers, illustrators, and other creators with paid projects in the AI economy. Pulse 2.0 interviewed Wirestock Co-Founder and CEO Mikayel Khachatryan to learn more.
Mikayel Khachatryan’s Background
When asked about his background and the experiences that led him to Wirestock, Khachatryan shared:
Before founding Wirestock in 2019, I was involved with several photo editing and social products, focusing on monetization and product development, which allowed me to develop a deep understanding of how creative tools scale and how creator economies function. That combination of economic vision and product experience is what led me to see the structural opportunity at the intersection of the creator economy and AI data demand. Not just as a marketplace problem, but as an infrastructure problem worth solving from the ground up.
This inspired me to found Wirestock. Wirestock builds premium, human-crafted visual training data for foundation model development, spanning video, image, 3D modalities, and more, including Pretraining, SFT, RL, and RLHF pipelines. Our global network of contributors in photography, videography, illustration, graphic design, 3D, and other creative disciplines earn consistent, predictable income through the paid creative projects that power this work.
How Wirestock Started
When discussing how the idea for Wirestock came together, Khachatryan explained:
We wanted to help content creators monetize their work more effectively. We quickly learned that there were many technical challenges in the process of licensing visual content and started building tools to address these challenges. That’s how Wirestock was born.
A Memorable Creator Story
When asked about his favorite memory from working at Wirestock, Khachatryan recalled:
One of our best-selling photographers was famous for family and country lifestyle photography. He earned and saved a considerable amount of money through Wirestock and used his Wirestock earnings to adopt a child. He then uploaded a new family photo with the adopted child, and that photo became one of the best-selling photos on Wirestock.
Core Products And Features
When describing Wirestock’s core products and capabilities, Khachatryan detailed:
Wirestock’s core product is an advanced multimodal visual training data platform built for AI research and development teams. We produce rights-cleared datasets engineered to specification across every stage of the model training pipeline, from large-scale pretraining corpora to supervised fine-tuning data, preference datasets, and reinforcement learning rubrics and verifiers. Our content spans photography, videography, illustration, 3D, graphic design, music, and other creative disciplines, giving labs a single sourcing partner for visual training data across modalities. Physical AI and world models are also areas of focus. We custom-build datasets for the top labs in the world to meet their specifications and also offer an off-the-shelf catalog of curated datasets.
On the creator side, we are building a platform that gives professional photographers, videographers, designers, illustrators, and other creatives access to consistent, paid project work in the AI economy. Creators are matched to projects based on their specific skills and experience, can track all earnings in one place, and receive reliable monthly payouts. As traditional stock income has become harder to sustain, Wirestock gives creators a direct path to scalable income from work that is genuinely valued. Their human expertise, trained eye, and creative range are exactly what make our datasets worth building.
The two sides reinforce each other. The depth and quality of our creator network are what make it possible to deliver advanced, specification-grade datasets at scale. And the demand from labs is what makes Wirestock a viable, growing income source for creators.
Navigating Constantly Evolving AI Market
When asked about recent challenges in the AI training data sector and how Wirestock has addressed them, Khachatryan observed:
AI labs’ data needs shift constantly, and staying useful to them means shifting with them. That’s the core of what we do at Wirestock: tracking where frontier development is headed and building the systems to produce data that keeps pace with it, rather than data built for yesterday’s models. Versatility isn’t a nice to have here, it’s the actual job. We stay close to what labs are working on, adapt quickly when priorities change, and keep pushing our production capabilities so the data we create moves the frontier forward instead of trailing behind it.
We are constantly learning new AI data training and fine-tuning methodologies to help our customers source and curate the best training data for their models. We then create processes that enable content creators to contribute their work and advance AI. It’s a continuous learning process that’s very rewarding and exciting.
Technology Evolution
When discussing how Wirestock’s technology has evolved since its launch, Khachatryan described:
We went from helping artists sell photos and videos in stock content marketplaces to curating and processing multimodal data for AI labs. What hasn’t changed is our focus on working with content creators and curating and annotating visual assets at scale.
Major Company Milestones
When asked about Wirestock’s most significant milestones, Khachatryan highlighted:
- Crossing $15 million in creator payouts.
- Partnering with six of the largest foundation model builders in the world.
Customer Success Story
When invited to share a customer success story, Khachatryan recounted:
One of the most meaningful validations of what we do came when a customer we had been supplying training data to launched a new version of its image generation model. That release quickly became recognized as one of the leading image generation models in the industry, outperforming competitors across the standard quality and alignment benchmarks that research teams use to evaluate these systems. We cannot name the customer, but the result speaks to something we believe deeply: the quality, structure, and diversity of training data are among the most consequential variables in what a model ultimately becomes.
What made that outcome particularly significant for us is that the datasets we built for that project were not off the shelf. They were engineered to specification, with custom content categories, precise annotation schemas, and modality pairings developed in close collaboration with the customer’s research team.
Funding And Revenue Metrics
When asked about Wirestock’s funding and revenue metrics, Khachatryan disclosed:
TechCrunch article for reference:
- $23 million Series A announced in May 2026.
- $26 million in total capital raised.
- $40 million run rate.
Market Opportunity
When discussing the total addressable market Wirestock is pursuing, Khachatryan estimated:
The total addressable market is $5-8 billion.
Wirestock’s Differentiators
When asked what differentiates Wirestock from its competition, Khachatryan emphasized:
- Compared with other data providers, we offer custom-engineered datasets built to exact specifications at every pipeline stage, rather than prepackaged catalogs or aggregated secondhand assets.
- Compared with stock marketplaces, our content is engineered for AI training rather than retrofitted from stock content.
- Compared with in-house data teams, we provide faster time to data with greater scale and diversity. Researchers can focus on training and evaluation rather than sourcing, annotation, and pipeline operations.
Future Goals
When discussing Wirestock’s future goals, Khachatryan concluded:
We want to continue pushing the boundaries of multimodal generative models with highly curated training data. We aim to partner with all the leading frontier model developers and become the top destination for multimodal and creative data.
