Agentic AI, Freshness, and the New Retrieval Layer: Key Insights from Bings Krishna Madhavan at Pubcon Cyber Week

Krishna Madhavan from Microsoft Bing joined Pubcon Cyber Week with a clear message for search professionals. AI agents are reshaping how information is fetched, ranked, and acted on. His talk outlined where the shift is happening and what marketers need to adjust before 2026 lands.

He opened with a simple line that anchored the entire session:

“AI agents are turning the web from a document system into an execution layer.”

AI Search Agents no longer behave like traditional search crawlers. They treat the web as simply a task environment, pulling in only the pieces needed to complete the actions for the user. This pushes retrieval, structure, and most-importantly, freshness to the center of modern visibility.

Krishna stressed that traditional SEO still carries weight, noting that: “retrieval and ranking are still fundamental to both workflows” whether the surface looks like search or an LLM result. The difference is that agents determine what to pull based on context, user preferences, and machine-readable signals, not just queries.

Freshness is Top Priority

“Stale or missing content will constrain the amount of retrieval we can do and push agents toward alternative sources.”

AI systems depend heavily on updated, structured content because grounding is now the bridge between live web data and model reasoning. IndexNow featured prominently as one of Bing’s most effective tools for this. He pointed out that:

“…half of clicked newly indexed URLs in the Bing SERP now come from IndexNow.”

It is not a ranking boost, but it is a freshness advantage that shapes what AI agents can see and take advantage of.

Another point that resonated with the Cyber Week audience was his emphasis on depth. “Depth matters way more than breadth.” Agents evaluate content through structure, clarity, and semantic density. They prioritize sources that present expertise in a compact, easy-to-parse for machines  format. Clean HTML, focused topical authority, and clearly segmented content blocks all increase visibility across agent-led surfaces and experiences.

 

Schema Schema Everywhere

Schema was framed as a practical requirement rather than a nice to have. With more AI systems consuming structured data directly through grounding calls, schema reduces ambiguity and gives agents faster pathways to trusted facts.

“Schemas are super useful. They help the system discern exactly what your information is without us having to guess.”

The talk closed on governance and sustained content health.
Marketers were encouraged to treat content as a living system. Updated text, clean markup, updated outlinks, and removal of dead paths all influence agent retrieval behavior.

Pubcon Cyber Week continues to surface the shift toward agentic search, and Krishna’s session made it clear that SEO is not disappearing. It is moving underneath AI systems that rely on clean, structured, verified sources. Visibility now starts long before a query is typed.

For more on live video replays, visit Pubcon.com.