Cognitiv has been at the forefront of experimenting with large language models (LLMs) for years, and since March 2023, we have integrated OpenAI’s GPT technology into our contextual advertising solutions. This week, Cognitiv announced an expanded collaboration with OpenAI to support the explosive demand for ContextGPT™—our solution that uses LLMs to deliver privacy-safe, precise contextual advertising. With advertiser adoption of ContextGPT growing more than 10x in the last 12 months, this innovation is proving to be a game-changer for brands and agencies.

In this Q&A, I sit down with Jana Jakovljevic, SVP of Partnerships at Cognitiv, to discuss how deep learning delivers real-world value to advertisers, the evolution of ContextGPT, and what our expanded collaboration with OpenAI means for the future.

Tell us more about the collaboration between OpenAI and Cognitiv.

Cognitiv has long recognized that at times, context can be a stronger predictor of conversion than user behavior. Since March 2023, we have been leveraging OpenAI’s GPT LLM to improve contextual targeting and drive higher returns on ad spend for our clients.

The new enterprise agreement with OpenAI allows us to scale our capabilities significantly. Previously, we were constrained by rate limits, preventing us from classifying the full breadth of internet content. With more processing capacity, we can analyze and understand an even broader set of URLs in real time. This ensures that our ContextGPT product provides highly relevant and brand-aligned ad placements at an unprecedented scale.

How is Cognitiv’s use case for GPT different from how others in the space approach contextual advertising?

Many contextual advertising providers still rely on outdated keyword-based approaches. These methods often lead to inefficiencies—either blocking too much content or failing to filter inappropriate placements. A well-known example is brand safety tools mistakenly blocking a Time magazine article about Taylor Swift due to simplistic keyword matching.

Cognitiv’s approach, powered by LLMs, offers a human-like understanding of content. Our system goes beyond surface-level keywords to grasp the full meaning and sentiment of a page. For example, if an article is about basketball, our AI can determine the exact content of the piece—whether it is a product review, an opinion piece, or a deep dive into sports analytics— and provide advertisers with precise targeting that aligns with their brand values.

For a deeper dive into the inefficiencies of keyword-based contextual advertising,  you can check out our Chief Technology Officer, Marc Hudasco’s LinkedIn article.

What types of results and use cases are you seeing with clients?

Our AI-driven contextual solutions are proving particularly valuable for clients looking to target niche or hard-to-reach audiences. For example, many brands struggle to engage Gen Z consumers, who have fragmented online behaviors. By analyzing the content they engage with—rather than relying on third-party data—our models help brands connect with this audience in a privacy-compliant way.

Also, clients without robust first-party data can still find their ideal customers by inputting simple prompts into our system. For example, a brand marketing ski equipment can identify audiences engaging with content about high-tech snow gear, winter travel, and outdoor sports.

Unlike off-the-shelf contextual segments, our models allow advertisers to define unique audience parameters based on the content they’re reading, which in turn delivers consumer intent.

How does Cognitiv’s technology address brand safety and sentiment analysis?

A major advantage of LLMs is their ability to assess sentiment and inclusivity in content. Traditional keyword-based filters often over-block or misclassify content, limiting advertisers' reach. 

Many advertisers avoid news content due to concerns about brand safety. However, not all news is negative. Our AI enables advertisers to carve out positive or neutral sentiment news with high precision, rather than excluding the entire category.

With our inclusivity filters, our models can identify and exclude content that may contain biased or exclusionary language. For example, a brand could avoid placements alongside content that reinforces harmful stereotypes. This nuanced approach allows brands to advertise responsibly while maintaining reach and relevance.

What’s on the roadmap for ContextGPT in 2025?

In 2025, ContextGPT is evolving beyond just understanding language—we are going multimodal. We have already demonstrated the ability of embeddings to capture nuanced meaning across more than ten languages. Now, we are expanding our capabilities to interpret images, video, and layout, offering a deeper, more contextual understanding of content and engagement patterns. Rather than simply analyzing keywords on a webpage, our multimodal models can process video, images, and layout to accurately determine the content’s true context. This not only ensures that ads are placed in the most relevant environments but also helps avoid harmful or misleading content by analyzing both text and visuals together.

Another key focus for 2025 is audience intelligence. We aim to deepen our understanding of how users engage with content, which will enable us to deliver smarter recommendations, richer insights, and, ultimately, more effective campaigns for brands.

As brands navigate a rapidly evolving digital landscape, advanced contextual intelligence is becoming more critical than ever. With the expanded collaboration between Cognitiv and OpenAI, ContextGPT is setting a new standard for contextual advertising—helping brands and agencies achieve unparalleled precision, privacy-safe targeting, and performance optimization at scale. 

Want to learn more about how ContextGPT can transform your advertising strategy? Reach out to our team at sales@cognitiv.ai and start the conversation today.

‍