Microsoft described Cosmos DB vNext emulator updates for AI coding agents on October 15. Its report covers October releases’ binary-protocol support, stricter indexing-metadata validation and experimental Nacelle gateway, alongside tools for loading test data and collecting traces.
Microsoft described vector-store support in Agent Framework for Python on October 15. Building on work transferred from Semantic Kernel, the system combines typed records, batch storage and retrieval, embedding generation, search filters and context providers exposing search tools to agents.
The sample creates PowerPoint presentations from financial and compliance documents. In AWS’s internal test, it produced six slides from the 90-page Amazon 2024 Annual Report in 70–90 seconds.
The AWS Marketplace integration turns HTTP 402 responses into x402 payments. AgentCore Payments manages wallet connectivity and enforces spending limits at the infrastructure layer.
The guide shows how to activate License Manager grants with a one-time script or event-driven CDK automation. Negotiated rates apply after a grant is activated.
AWS detailed GPU kernel optimizations for mixture-of-experts models on SageMaker AI on October 13. The company described swapping matrix operands, routing FP8 operations to the Triton backend and removing unnecessary just-in-time recompilations.
AWS Artificial Intelligence Blog1 min read Data included
Microsoft Research presented Magnetar on October 15, a system producing zero-knowledge proofs for programs written in Rust, Go or C and compiled for RISC-V. The research aims to let others verify execution without revealing inputs or rerunning the computation.
Microsoft AI said on October 15 that it completed pre-training MAI-Base-2, a mixture-of-experts model with 2.8 trillion total and 80 billion active parameters. The company says training used 16,000 GB200 GPUs over 38 days in August and September; the model is still undergoing post-training.