As AI agents become more adept at reasoning and autonomously handling long-term tasks, ensuring they have access to the necessary information poses a significant engineering challenge. The industry’s focus is shifting from pre-loading context to allowing agents to dynamically retrieve and navigate data. Redis tackles context management with a context engine architecture based on four key aspects: on-demand context retrieval, constantly updated data, quick access, and a progressively improving memory layer. This involves creating materialized data views with a semantic layer, rather than providing direct access to production databases, with a memory system that compacts and extracts information asynchronously.
Simba Khadder, leading AI strategy at Redis and former co-founder of FeatureForm (acquired by Redis in 2025), discusses with Kevin Ball why context is a crucial challenge in agentic AI, the differences between context engines and traditional RAG architectures, the role of materialized views in reliable data pipelines, improvements in memory systems via asynchronous processes, and how engineering teams need to adjust their approaches in the rapidly advancing AI-driven development landscape.
Note: This episode is sponsored by Redis. Kevin Ball is the vice president of engineering at Mento and offers coaching for engineers and leaders. He co-founded two companies, initiated the San Diego JavaScript meetup, and leads the AI inaction group via Latent Space.
Please click [here](http://softwareengineeringdaily.com/wp-content/uploads/2026/04/SED1923-Redis.txt) to see the transcript of this episode.