Workday built an internal agent developer platform on LangGraph for orchestration and LangSmith for tracing, observability and evals, running production AI agents on its own infrastructure. A four-person team delivered it under some of the strictest security, audit and data governance constraints in enterprise software, targeting agents in production by end-2026.
Agent Orchestration
智能体编排
Engineering pattern for scheduling, monitoring and steering multiple parallel agents via a unified interface.
Curated Intelligence
Recalc is an open source, locally run spreadsheet in which every cell can act as its own terminal, wiring directly into coding agents such as Claude Code and Codex so users can watch agents fill cells in real time and steer them by prompting or editing outputs. The project is in early beta, supports macOS and experimental Linux, and has not been tested on Windows.
AI video generation is fast, but creators often need roughly ten attempts to get one satisfactory clip, and this repeated trial-and-error drives model-call, screening and rework costs. Tencent Cloud and peers point to a missing structured layer between scripts and generation models, with shot control, character consistency and editability remaining the main barriers to commercial delivery; the industry is turning to agent orchestration to connect generation, processing and revision.
Angel investor Greg Isenberg summarized the current startup playbook in one chart: acquire valuable proprietary context, orchestrate AI agents to build on it, feed the output back into the loop, and call humans like an API. He argues the company with the best context and fastest loop wins. The claim is experiential, with no data or verifiable cases attached.
Silicon Valley startup AFK AI says its product Comma already implements a single-router, multi-worker multi-agent architecture, using continuous context and long-term memory to keep agents running 24/7 across many parallel tasks, with code open-sourced. The piece frames this as converging with DeepSeek Harness's Super-agent vision around human attention as the bottleneck, but offers mostly conceptual narrative without verifiable architecture details or benchmark results.
Security researchers found hundreds of instances of Artex, an open-source autonomous penetration-testing tool, publicly exposed on the internet and accessible to anyone, potentially enabling cyberattacks. Built on AI agents that automate the attack chain from reconnaissance to data exfiltration, Artex has been linked to multiple attacks since its July release. It has since gone closed-source, but many usable code copies remain on code-sharing sites.
Nvidia published benchmark results for its Vera CPU on agentic workloads, claiming that in a 100 MW data center scenario it supports 1.6x more agents per watt than the fastest x86 CPU and delivers 2x agent speed versus the most efficient x86 part. Early partners including Perplexity, DeepInfra, ClickHouse, Daytona and Modal reported speedups in coding, orchestration and concurrent sandbox scenarios, reframing the CPU side as a scaling bottleneck for agent deployments.
Meta is acquiring Conductor, an AI coding startup whose product lets developers orchestrate multiple AI coding agents simultaneously; the price was not disclosed. The deal fits a broader push by major AI labs to buy model-adjacent tooling and lock in developers, with OpenAI announcing a Pierre Computer acquisition the same day. Conductor Cloud shuts down on November 20, and the desktop app stops receiving new models in early 2027.
Meta has confirmed its acquisition of Conductor, a 2024-founded AI coding startup whose tool orchestrates AI agents across multiple parallel, isolated workspaces. The founder will join Meta Superintelligence Labs, while Conductor's cloud service shuts down on November 20 and core features continue until early next year. The deal marks another step in Meta's push to catch up in agentic coding.
Voyager has launched a desktop AI video agent that bridges generative video models like Seedance 2.5 with professional editing software such as Blender and DaVinci Resolve. Operating like a coding agent, it plans shots before invoking models, reportedly reducing API calls by 2-3x and cutting costs in half.