PicsArt’s creative AI playbook: A vision for contextual intelligence, AI agents

Whether you are an Android or iOS person, most people have heard of it ImagesArt. The platform was launched over a decade ago and has become one of the go-to services for all things image and video editing, with more than 150 million monthly active users.

However, it has not been an easy journey for the company. Despite being a pioneer in smartphone-based editing, the company has seen significant competition from players like Canva and Adobe, who have been playing a cat-and-mouse game for some time and building their own similar products. When I recently spoke with Artavazd Mehrabyan, the company’s CTO WCIT Conference in Armenia he was quite vocal about the challenges, saying that it is difficult to be different or at least stay different for long in this market.

“Many things that PicsArt had before were copied to the competitors. PicsArt was the first all-in-one editing service on mobile. Before 2011, there was no other player. We started this approach and it was copied, among other things,” said Mehrabyan. He pointed out that the same thing is happening with AI, where competitors, including mainstream photo services, are offering very similar capabilities.

For example, PicsArt offers object generation, allowing users to use advanced AI to create the required photo elements. The same capability is also included in other products in the category, creating something of an overlap.

But instead of trying to differentiate itself by adding more tools to its existing suite of more than two dozen AI capabilities, the company wants to make a mark on users by improving the quality of what it delivers. Specifically, Mehrabyan says, the focus is on how they productize and tailor features to help customers achieve their goals – whether they want to remove a specific object from a vacation photo or generate visually appealing ads complete with images and text .

Training high-quality creative AI

In the early stages, when AI didn’t exist, Mehrabyan said most of PicsArt’s technology research and efforts were focused on making mobile editing seamless.

“It was very difficult to get all this editing functionality working offline on the device. Then the next challenge was scaling our ecosystem and infrastructure to support a growing user base. This brought us to hybrid infrastructure. We started with multi-cloud and a data center, which is still the best solution so far because it is more cost-efficient, better performing and very flexible,” explains Mehrabyan.

With this tech stack in place, the company launched its first AI feature in 2016, running a number of small models offline on user devices. This gradually turned into a large-scale AI effort, with the company transforming into an AI-first organization and leveraging its infrastructure and backend services to serve larger models and APIs for more enhanced capabilities, such as background removal/replacement . More recently, as the generative AI wave took shape, PicsArt began training its own creative AI models from scratch.

In the creative domain it is very easy to lose a user. A small mistake here or there (leading to low-quality results) and chances are the person won’t come back. To prevent this, PicsArt is extremely focused on the data side. It selectively uses data from its own network – marked by users as public and freely editable – to train its AI models.

“We have a special ‘free to edit’ license. If you post your image publicly and label your image – from a stock photo in any category to a sticker or background – as free to edit, another user of the service can reuse the image or work on top of it. So essentially the user contributes this image to the community and PicsArt itself,” said Mehrabyan.

The license has existed since the service’s inception and has given PicsArt a huge inventory of user-generated content for training AI. But as the CTO noted, not everything is high quality and ready to use. The data must go through multiple layers of cleaning and processing, from manual to AI-driven, to be transformed into a secure, training-ready dataset.

“We end up with quite a large dataset owned by PicsArt. We don’t need any additional data,” he said.

However, having a large amount of high-quality data was only part of the puzzle.

The real challenge for PicsArt was, as Mehrabyan described, building the “data flywheel.” A self-reinforcing cycle that includes not only data accessibility, but also aspects such as how to annotate data, how to use it and ultimately how to use it as part of a continuous learning process to improve over time .

Setting up a feedback loop to achieve this was a long and complex process, he said.

“We built our own annotation technology. We have internally developed all related infrastructure and ecosystem technologies, including technologies for identifying and classifying images, tagging them and adding different types of labels,” said Mehrabyan. “We then built a team to help refine the pipeline and provide feedback over time. It has largely been very automatic, AI-driven with human feedback in between, so we can continuously improve.”

Feedback loop leads to contextual intelligence

While the human-driven feedback loop has been a crucial part of improving PicsArt’s products—improving the quality of the output they generate—it’s also moving the company toward what Mehrabyan calls “contextual intelligence,” or the platform’s ability to understand user needs. and deliver exactly what they want.

This feature is especially important for the platform’s growing group of business users who want to do their work directly on their smartphones. Whether that is generating graphics or a full-fledged advertisement for a social media campaign. The platform is still mainly used by people who want to edit personal content, but the company says research shows many want to put it to work, especially for marketing purposes.

“Contextual intelligence not only tracks your history or what you were doing to help you be more productive on your journey, but also predicts your next intentions. It is both reactive and proactive,” he explains.

This way, every time an individual uses the platform to create something for their work, they don’t have to define brand language and tonality. The product would already have context and use that to generate the required content. Mehrabyan said the company also plans to release a brand kit that will allow users to customize this context to their needs and further improve the quality of generations.

Creative AI agents on the move

Mehrabyan says contextual intelligence will ultimately lead PicsArt to an agent-based ecosystem. Here users will have a kind of co-pilot – with all the relevant knowledge about their work and design preferences – to help them with their tasks.

“This co-pilot understands your intentions and historical context to provide interactive support and guide you to be even more productive. We see this use case as integrated across the entire PicsArt ecosystem, from the user’s perspective,” he said.

In addition, he also expects AI agents to help PicsArt users perform some tasks in bulk. For example, if a user needs to apply the same design or logic to different resources, they can use an agent to automate the workflow on their behalf.

In this way, the company hopes to be a major force in the creative industry, staying ahead of the competition and allowing users to grow their creativity and ultimately their business, without too much effort.

Mehrabyan noted that AI will bring a big change, but users – from businesses to designers and marketers – should try to understand how it affects them and take advantage of the changes to do more than is currently possible.

“From the current perspective, this will have a negative impact. But if you look at the perspective from another side, for example from the future, you will see that those people will use AI to learn much more. They will no longer be limited specialists. They will cover broader areas deeper and faster with the help of AI,” he noted.

According to Future market insightsThe global AI image editor market is expected to grow from $80.3 million in 2024 to $217.9 million in 2034, with a CAGR of 10.5%. Meanwhile, AI-driven generation, which has become a core part of most image editing tools/services including PicsArt, expected to grow 38% from $8.7 billion in 2024 to $60.8 billion in 2030.