PubMatic develops 47 agentic AI projects in 12th annual hackathon
PubMatic held its 12th annual hackathon with 416 participants across 5+ countries, developing 111 projects focused on agentic AI. The event highlighted a shift from AI that merely responds to AI that can autonomously plan, execute, and adapt, moving beyond assistive features into active execution within their platform. PubMatic engineers are increasingly integrating agentic AI into their development processes for continuous optimization and productization of operational AI capabilities.
Key Takeaways
- 47 out of 111 hackathon projects (42%) focused on agentic AI using LLM reasoning and Model Context Protocol (MCP) workflows.
- 54% of participating teams were cross-functional, while 14 teams comprised entirely non-engineers building end-to-end AI applications.
- New systems prioritize continuous real-time optimization of pricing, yield, and targeting over traditional periodic manual tuning.
- Engineering teams utilized Model Context Protocol (MCP) to ensure interoperability and open communication across agentic prototypes.
Why It Matters
The transition from assistive to agentic AI indicates a maturation of the ad tech stack, where autonomous agents manage high-frequency tasks like bid optimization and diagnostic troubleshooting. By embedding these capabilities directly into the platform, PubMatic reduces the operational latency typically associated with manual programmatic management. This shift reflects a broader industry trend toward 'AgenticOS' models, where interoperable protocols like MCP allow multi-agent collaboration across fragmented video ecosystems. To gauge future scale, watch for the speed at which these internal prototypes transition into customer-facing products, potentially driving the 80% growth currently seen in PubMatic's emerging AI revenue streams.
Additional Context
The hackathon findings coincide with significant commercial momentum for PubMatic’s agentic infrastructure. In June 2026, the company launched its first fully agentic CTV campaign in Spain with Havas and Telefónica Group’s Movistar. The activation used AgenticOS to autonomously ingest natural language briefs, select inventory, and apply frequency controls, resulting in an 18% lower CPM compared to target benchmarks. This regional expansion follows similar deployments in France and the Netherlands, reinforcing a shift toward autonomous media buying in premium video environments. Institutional support for agentic standards is accelerating across the vendor landscape. Per AdRoll and GlobeNewswire in April 2026, PubMatic and AdRoll announced a functional integration using the Model Context Protocol (MCP) to solve deal underdelivery diagnostics. This allows demand-side agents to query supply-side data in real time, bypassing traditional manual support tickets. This architectural move aligns with a 2026 IAB UK report stating that 74% of industry members now use agentic AI, though only 4% consider themselves 'agent-first.' Financially, PubMatic is successfully trading on its AI-first transformation. In its Q1 2026 earnings report from May, the company beat revenue estimates at $62.6 million and reported processing 94.2 trillion ad impressions. Management noted that emerging revenues from AI-powered solutions grew 80% year-over-year, now representing 14% of total revenue. Per Investing.com, the company’s ability to scale impression volume by 26% while keeping cost of revenue flat demonstrates the operational leverage gained from automating internal engineering and diagnostic workflows through its agentic stack.
Read full article at pubmatic.com
Get this in your inbox → Subscribe
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source