Most people have heard of whether you are an Android or iOS user PhotosArt. The platform launched over a decade ago and has turn into one of the hottest photo and video editing services, with over 150 million monthly lively users.
However, it was not an easy path for the company. Despite being a pioneer in smartphone editing, the company has seen significant competition from players resembling Canva and Adobe, who have been playing a cat-and-mouse game for quite some time, creating their very own similar products. When I last spoke with Artavazd Mehrabyan, the company’s CTO WCIT conference in Armenia, he has been quite vocal about the challenges, saying that it is difficult to be or at least remain different for a very long time in this market.
“A lot of what PicsArt had before was copied by competitors. PicsArt is the first comprehensive editing service for mobile devices. Before 2011, there was no other player. We started with this approach and it was copied, among others,” Mehrabyan said. He identified that a similar thing happens with AI, where competitors, including mainstream photography services, offer very similar possibilities.
For example, PicsArt offers object generation, allowing users to make use of advanced artificial intelligence to create the required photo elements. The same functionality is also used in other products in this category, creating some overlap in functionality.
But fairly than differentiate itself by adding more tools to its existing pool of greater than two dozen AI features, the company desires to impress users by improving the quality of what it delivers. In particular, Mehrabyan said, the focus is on how one can create and customize features to assist customers achieve their goal – whether or not they wish to remove a specific object from a vacation photo or generate visually appealing ads complete with photos and copy.
High-quality creative AI training
Mehrabyan said that in the early stages, when AI was not yet commonplace, most of PicsArt’s research and technology efforts focused on ensuring seamless editing on mobile devices.
“It was very difficult to run all these editing features on the device while offline. The next challenge was scaling our ecosystem and infrastructure to support our growing user base. This led us to hybrid infrastructure. We started with multi-cloud and data center, which remains the best solution so far because it is more cost-effective, highly efficient and very flexible,” explained Mehrabyan.
With this technology stack, the company launched its first AI feature in 2016, running several small models offline on users’ devices. This regularly evolved into a large-scale AI effort where the company evolved into an AI-first organization and leveraged its infrastructure and backend services to support larger models and APIs for more enhanced capabilities , resembling removing/replacing background. More recently, as the generative AI wave takes shape, PicsArt began training its own creative AI models from scratch.
In the creative domain, it is very easy to lose users. A small mistake here or there (resulting in low-quality results) and there is a good probability that the person is not going to come back again. To prevent this from happening, PicsArt is extremely focused on the data side. It selectively uses data from its own network – marked by users as public and edit-free – to coach artificial intelligence models.
“We have a special license for the free edition. If you post publicly and mark your image – from a stock photo in any category to a sticker or background – as free to edit, this allows another user of the service to reuse or work on it. Basically, the user shares this image with the community and with PicsArt itself,” Mehrabyan said.
The license has been in effect since the service’s inception and has provided PicsArt with a vast resource of user-generated content for AI training. However, as the CTO noted, not every thing is prime quality and ready for immediate use. Data must go through multiple layers of cleansing and processing, from manual to AI-based, to be transformed into a secure, training-ready dataset.
“At the end of the day, we have quite a large data set owned by PicsArt. We don’t need additional data,” he said.
However, having loads of high-quality data is only one a part of the puzzle.
As Mehrabyan described, the real challenge for PicsArt was building a “data flywheel.” A self-reinforcing cycle that features not only the availability of information, but also elements resembling how one can describe the data, how one can use it, and ultimately how one can use it in a continuous learning process to enhance over time.
He said creating a feedback loop to realize this goal was a long and complex process.
“We created our own annotation technology. We have internally developed all related infrastructure and ecosystem technologies, including technologies to identify and classify images, tag them and add various types of labels to them,” Mehrabyan said. “We then formed a team to help refine the pipeline and provide feedback over time. This was largely automated, AI-powered and augmented by human feedback, allowing us to continually improve.”
The feedback loop results in contextual intelligence
While the human-led feedback loop has played a key role in improving PicsArt’s products – improving the quality of the results they generate – it is also leading the company to what Mehrabyan calls “contextual intelligence,” the platform’s ability to grasp users’ needs and deliver exactly what they need.
This feature is especially vital for the platform’s growing base of business-focused users who wish to get work done directly from their smartphones. Whether it’s generating graphics or full-fledged promoting for a social media campaign. The platform is still mainly used by people seeking to edit personal content, but the company says research shows many people wish to use it for work, particularly for marketing purposes.
“Contextual intelligence not only tracks your history and what you’ve been doing to help you be more productive on your journey, but it also predicts your next intentions. This is both a reactive and proactive approach,” he explained.
This way, every time a person uses the platform to create something for their work, they will not have to define the language and tone of the brand. The product would already have context and use it to generate the required content. Mehrabyan said the company also plans to release a brand toolkit that can enable users to tailor this context to their needs and further improve the generation experience.
Creative AI agents on the way
Ultimately, Mehrabyan says, contextual intelligence will lead PicsArt to an agent-based ecosystem. Here, users will have a co-pilot of sorts – with all the essential knowledge about their work and design preferences – to assist them complete their tasks.
“This co-pilot will understand your intentions and historical context to provide interactive support and guide you to even greater productivity. From a user perspective, we see this use case as integrated into the entire PicsArt ecosystem,” he said.
In addition, it also expects AI agents to assist PicsArt users perform certain tasks collectively. For example, if a user needs to use the same design or design logic to several resources, they’ll use an agent to automate the workflow on their behalf.
In this manner, the company hopes to turn into a key driver of the creative industry, staying ahead of the competition and allowing users to develop their creativity, and ultimately their business, without much effort.
Mehrabyan noted that AI will bring about major changes, but users – from businesses to designers and marketers – must try to grasp the way it affects them and use the changes to do greater than is currently possible.
“From the current point of view, it will have a negative impact. But if you look at it from another angle, perhaps from the future, you will see that these people will use artificial intelligence to learn much more. They will no longer be narrow specialists. Thanks to artificial intelligence, they will cover wider areas deeper and faster,” he noted.
According to Insights into future marketsThe global AI image editor market is projected to grow from $80.3 million in 2024 to $217.9 million by 2034, at a CAGR of 10.5%. Meanwhile, the AI-powered generation that has turn into a core a part of most image editing tools/services, including PicsArt, is is estimated to extend 38% from $8.7 billion in 2024 to $60.8 billion in 2030.