AWS published Part 2 of its series on governing AI agent access using Okta Cross App Access with Amazon Bedrock AgentCore. The post moves from Part 1's single-agent, two-MCP-server proof of concept to a topology that supports hundreds of agents and thousands of tools, shifting authorization checks out of individual MCP servers into a centralized gateway.
Agent Runtime
智能体运行时
Infrastructure layer responsible for agent scheduling, lifecycle management, and resource allocation.
Curated Intelligence
AICon Beijing will run Dec 18-19 with 13 tracks spanning agent engineering, inference optimization, and agent evaluation and safety. Tencent engineer Chen Siying will present an AI-native enterprise dev toolchain in the Agentic Software Engineering track, covering tool protocolization, agent runtime engineering, context engineering platforms, and a friction-signal feedback loop. The piece is primarily an agenda announcement without concrete technical parameters or verifiable results.
At its 2026 Digital Ecosystem Summit, Tencent Cloud framed enterprise agent adoption as a connectivity problem and rolled out 40-plus product capabilities across office collaboration, enterprise data, permission governance and runtime infrastructure, including WorkBuddy Enterprise, ADP, AgentHosting, Agent Runtime and TokenHub. The announcement is a broad vendor showcase rather than a technical disclosure, with few concrete specs or verifiable benchmarks, so its value lies mainly in mapping Tencent's agent product portfolio.
Octomind is released as a Rust-based, CLI-first AI coding agent runtime that grants LLMs real access to repositories via MCP tools for file operations and shell execution. It supports multiple modes including interactive, piped stdin, background daemons, and WebSocket, featuring built-in hooks and validators to replace manual approvals in autonomous workflows.
Tencent Cloud announced a major upgrade to its MaaS product line at the Global Digital Ecosystem Conference, shifting focus from simple API calls to controllable task delivery via components like TokenHub, TI-ONE, and Agent Runtime. This move addresses enterprise challenges in multi-model scheduling, governance, and cost accounting, aiming to push AI applications into production environments.
CITIC Securities reports that DeepSeek's open-source agent runtime framework, Harness, is pivotal for its transition from model provider to platform ecosystem. The framework leverages plugin architecture and cross-agent collaboration to convert engineering advantages into commercial revenue and user stickiness.
DeepSeek has consolidated its Harness, Web, App, and Platform teams into a unified Product Division led by Cui Tianyi. This move aims to streamline development from the underlying Agent runtime framework to upper-layer applications, enhancing internal synergy.
Microsoft CEO Satya Nadella published a manifesto treating AI models as high-privilege 'insiders,' advocating for the separation of intelligence provision from control. He outlined seven core security principles—including model diversity, full observability, and independent auditing—emphasizing that enterprises must implement externalized permission management rather than relying on vendor guarantees.
The 2026 QCon Shanghai conference is scheduled for October, focusing on AI Native architecture and Agent Runtime. A Kuaishou expert will present on 'Conan AI Agent' practices for mobile stability assurance. This is primarily a conference announcement with limited technical details available prior to the event.
The 2026 QCon Shanghai conference announces its October schedule, focusing on the paradigm shift from 'building AI' to 'harnessing AI'. It covers nine key areas including Agent Runtime and Loop Engineering, addressing the challenge of ensuring correctness in large-scale engineering despite reduced code generation costs.