AI Agents Get Smarter, Faster, and Cheaper

OpenAI is enhancing its AI offerings with several key updates, including improvements to its ChatGPT deep research tool and its Responses API, while the industry explores new memory architectures for AI agents. These advancements, announced on February 10, 2026, aim to improve the capabilities and efficiency of AI-driven applications.

ChatGPT's deep research tool is being updated with a full-screen viewer, allowing users to scroll through and navigate AI-generated reports, according to The Verge. This new feature includes a table of contents to jump to specific sections of the report.

Simultaneously, OpenAI is upgrading its Responses API, which allows developers to access multiple agentic tools like web search and file search with a single call. These updates include Server-side Compaction and Hosted Shell, signaling a shift away from limited AI agents, as reported by VentureBeat.

The industry is also exploring alternative memory architectures for AI agents. One such implementation is "observational memory," an open-source technology developed by Mastra. This technology prioritizes persistence and stability over dynamic retrieval, potentially reducing AI agent costs by a factor of 10 and outperforming RAG on long-context benchmarks, according to VentureBeat.

The advancements in AI are also impacting various sectors, including fraud protection. Mastercard's Decision Intelligence Pro (DI Pro) uses sophisticated AI models to analyze individual transactions and pinpoint suspicious ones in milliseconds, as stated by VentureBeat. This capability is crucial, given that Mastercard's network processes approximately 160 billion transactions annually, with peak periods experiencing surges of 70,000 transactions per second.

However, the widespread deployment of agentic AI is still evolving. While agentic AI was anticipated to transform Global Business Services (GBS) in 2025, the fundamentals required to scale have been missing, according to VentureBeat Contributing Editor Taryn Plumb, citing input from Google Cloud and Replit.