Regardless of whether you’re a die-hard Android fan or an ardent Apple enthusiast, the vast majority of people are familiar with the terms. The platform, which launched more than a decade ago, has evolved into a leading destination for all things image and video editing, boasting over 150 million monthly active users.
Despite appearances, the corporation’s path has not been without its challenges. Despite being a pioneer in the smartphone-based modification space, the company has faced significant competition from players such as and , who have been engaged in a prolonged game of technological one-upmanship – developing their own similar products. In a recent conversation with Artavazd Mehrabyan, Chief Technology Officer at the corporate, I gained valuable insights into the current landscape in Armenia. He candidly discussed the challenges, emphasizing that it’s crucial to stay ahead of the curve or remain distinct for an extended period in this market.
Various problems that PicsArt encountered previously have been replicated by its competitors. PicsArt emerged as the leading all-encompassing photo editing platform for mobile devices. Prior to 2011, there were no other participants. According to Mehrabyan, we initially employed a specific approach that ultimately became replicated alongside numerous other problems. He noted that a parallel phenomenon is occurring in AI, where rivals and mainstream image providers alike are offering remarkably similar features.
PicsArt’s object technology empowers users with advanced AI capabilities, enabling them to craft precise picture elements. The identical functionality has also been integrated into various products within the category, resulting in an overlap of product types.
Instead of relying solely on adding more tools to its arsenal, the company is endeavouring to leave a lasting impression on clients by elevating the quality of what it provides. Mehrabyan emphasized that the primary objective lies in productizing and tailoring the offerings to facilitate prospects’ journey to achieving their goals – regardless of whether they aim to extract specific objects from a travel snapshot or create visually captivating commercials replete with images and realistic replicas.
Coaching high-quality inventive AI
During the early stages of development, prior to the advent of AI as a significant factor, Mehrabyan focused primarily on optimizing PicsArt’s mobile-based editing capabilities.
“It took significant effort to integrate the necessary modifications into the system while it was offline.” As the user base grew exponentially, we faced the challenge of scaling our ecosystem and infrastructure to meet the demands of this burgeoning market. As a result, we transitioned seamlessly into a hybrid infrastructure. “We initially opted for a multi-cloud strategy, which remains the optimal solution due to its exceptional value proposition: it’s more cost-effective, highly performant, and incredibly flexible,” Mehrabyan explained.
With this technology stack in place, the corporation successfully launched its inaugural AI feature in 2016, deploying a series of compact models that operated independently offline on consumer devices. By transforming this step-by-step approach into a comprehensive AI initiative, the company refashioned itself as an AI-driven organization, harnessing its infrastructure and backend resources to deliver larger-scale models and APIs that unlock advanced features such as seamless image manipulation, including background removal and substitution capabilities. Recently, with the surge of generative AI technologies, PicsArt started training its own custom AI models from the ground up.
In the realm of innovation, it’s remarkably easy to misplace a customer. A minor mistake here and there can significantly increase the likelihood that the individual will not return to your business again? To proactively mitigate concerns, PicsArt places extraordinary emphasis on information-centric approaches. The platform leverages user-generated content within its open-source network, designated as publicly accessible and modifiable, to train AI models through collaborative learning.
“We have obtained a specialized ‘free to edit’ license.” When sharing public content and labeling images – whether taken in-house or sourced from a particular category to a sticker or backdrop – with permission to edit, this allows another user of the platform to repurpose or build upon it. Mehrabyan explained that consumers are essentially sharing a part of themselves with the community and PicsArt platform by uploading their images.
Since the inception of the service, the licence has enabled PicsArt to amass a substantial repository of user-generated content, providing a rich source of data for training artificial intelligence models. Despite this, not all components meet the standards for immediate deployment, as per the CTO’s assessment. Through a series of meticulously crafted layers, the information undergoes a thorough cleaning and processing regimen, guided by human oversight and augmented intelligence, ultimately transforming it into a robust, training-ready dataset that meets stringent security standards.
“At the conclusion of this project, we have successfully assembled a massive, proprietary dataset belonging exclusively to PicsArt.” He observed that we didn’t necessarily need additional insight.
Despite having a significant amount of high-quality information at one’s disposal, that was only one crucial element of the equation.
The challenge for PicsArt, as Mehrabyan outlined, was to create the “knowledge flywheel,” a self-sustaining cycle that integrated not only knowledge accessibility but also features such as annotating knowledge, utilizing it, and ultimately leveraging it as part of an ongoing learning process to enhance over time.
The process of setting up a suggestion loop proved to be an extended and complex endeavor, according to him.
We developed a proprietary annotation expertise. “We created all necessary infrastructure and ecosystem technologies in-house, including those for image recognition, classification, tagging, and attaching diverse labels,” Mehrabyan explained. We subsequently established a dedicated team to continuously optimize the pipeline, offering recommendations for future improvements. “It’s primarily been a highly automated process, driven by artificial intelligence and punctuated by human input to ensure steady progress.”
The suggestions loop enables organisations to generate contextual intelligence by leveraging the collective knowledge and insights of their stakeholders. By fostering open dialogue and encouraging diverse perspectives, companies can create an environment that fosters innovation, creativity, and continuous learning. This collaborative approach not only strengthens employee engagement but also drives business success through data-driven decision making and strategic alignment.
While human-driven suggestions have been instrumental in refining PicsArt’s offerings, improving output quality, this loop is also propelling the company towards “contextual intelligence” – the ability to grasp consumer needs and deliver tailored experiences precisely when they matter most.
This operation is particularly crucial for the platform’s growing contingent of business-oriented users seeking seamless execution of tasks on their mobile devices, thereby necessitating a user-friendly interface that seamlessly integrates with existing workflows. Regardless of whether it’s creating graphics or a comprehensive advertisement for a social media marketing endeavour. Although primarily used by individuals editing personal content, the company’s research suggests a significant desire to apply the platform in a professional setting, notably for advertising purposes.
Here is the rewritten text:
“Contextual intelligence goes beyond tracking your past actions, using your historical data to inform and predict your next steps, ultimately driving increased productivity throughout your journey.” He distinguished between reactive and proactive approaches, explaining that they are distinct.
When utilizing the platform to craft content for professional purposes, users won’t need to expend energy on crafting tone and language guidelines each time they post. The product utilizes existing context to produce the necessary content. Mehrabyan revealed plans for a bespoke module feature, enabling clients to customize and refine the output to suit their specific needs, thereby elevating the quality of generated content.
Artificial Intelligence brokers devise innovative strategies to optimize market performance?
Finally, Mehrabyan says that contextual intelligence will lead PicsArt to another. Customers can access a dedicated copilot portal featuring personalized profiles that consolidate their relevant work details, design specifications, and preferences to streamline task management and decision-making processes.
“This AI-powered copilot would intuitively grasp your intentions and historical context to provide personalized assistance, empowering you to work more efficiently.” From the consumer’s vantage point, we envision this application within the comprehensive PicsArt framework.
Moreover, he anticipates that AI-powered brokers will enable PicsArt users to automate and execute certain tasks in large quantities. When consumers require identical designs or logic from multiple sources, they often utilize an agent to streamline and automate the workflow process for their benefit.
By embracing innovation, the corporation aims to take the lead in its field, propelling itself ahead of competitors while empowering clients to unlock their creative potential, thereby boosting businesses with minimal exertion.
As artificial intelligence revolutionizes the landscape, Mehrabyan’s predictions indicate significant shifts ahead; meanwhile, stakeholders – ranging from corporations to creatives and innovators – must proactively grasp the implications and capitalize on the transformations to exceed current capabilities.
From our current viewpoint, it appears that it will likely have a detrimental impact. When examining the impact of AI from a distinct vantage point, such as a long-term perspective, it becomes clear that individuals will increasingly rely on artificial intelligence to facilitate deeper learning and exploration. Individuals will not be narrow experts. With AI’s aid, they will cover wider territories more profoundly and expediently.
According to industry forecasts, the global AI-powered picture editing market is anticipated to grow significantly over the next decade, increasing from approximately $80.3 million in 2024 to an estimated $217.9 million by 2034, with a compound annual growth rate (CAGR) of 10.5%. Meanwhile, AI-driven technology, now a core component of many image editing tools and companies, including PicsArt, is expected to grow significantly, expanding from approximately $8.7 billion in 2024 to an estimated $60.8 billion by 2030, representing a substantial increase of 38%.
