On Nov 5, 2026, the AI agent infrastructure sector saw a dense wave of launches. According to Product Hunt and Hacker News, at least five products or projects targeting agents' underlying needs launched that day, spanning orchestration and scheduling, memory management, cloud compute, device control and video collaboration. Most are at an early stage, but their concentrated debut shows developers are building around the agent runtime and tooling layer.
Orchestration and memory: Rivet open-sources actor orchestrator, Living Brain launches memory layer
Rivet (YC W23) posted a Launch HN on Hacker News, announcing the open-sourcing of an orchestrator built on the actor model for agent workloads. Co-founders Nathan and Nicholas said the approach does not run a heavyweight sandbox per agent; instead, each agent gets a lightweight actor backed by S3.
Actors provide a long-lived process for long agent turns, in-memory state for context, a WebSocket server for multiplayer, durable SQLite storage backed by S3, and hibernation when idle.
The Rivet team said the model is similar to Cloudflare Durable Objects but open-source and self-hostable. Its technical grounding comes from building infrastructure for multiplayer online games such as Diep and Krunker during YC in 2023. To cope with Postgres throughput bottlenecks at the time, the team wrote Lua scripts on Redis, then expanded into an orchestrator supporting lightweight workers that once served tens of millions of monthly active users. As agents became popular, the team concluded that sandboxes are expensive, idle most of the time and slow to start, while the actor model better suits long-running agent tasks.
On the same day, Living Brain launched on Product Hunt, positioned as an "Interoperable, efficient memory layer for agentic AI." The product aims to give agents memory management across systems and sessions, complementing Rivet's persistent state storage at a different layer.
Compute and devices: Maritime offers customizable cloud computers, Glasshand connects to real devices
Maritime went live on Product Hunt with the tagline "Cloud computers for any AI agent, customized your way." The product gives agents a configurable compute environment in the cloud, contrasting with Rivet's lightweight actor approach — the former emphasizes a full computer environment, the latter high-density lightweight processes.
Glasshand, meanwhile, proposes "Give your agent a real iPhone and Mac." The product lets agents directly control real Apple devices rather than simulated environments, targeting automation on mobile and desktop. Such approaches may involve device farms or remote-control technology, but the source material did not disclose the specific implementation.
Collaboration: Bridge explores video conferencing in the agent era
Bridge launched on Product Hunt, positioned as "Video conferencing for the agentic era." The product explores collaboration formats in which agents participate in or assist video meetings; specific feature details were not elaborated in the source material.
Industry watch: competition shifts to the runtime and tooling layer
These five threads appeared on the same day, covering the orchestration, memory, compute, devices and collaboration tools agents need to run. In its launch, Rivet said outright that sandbox approaches are "expensive, idle most of the time and slow to start," reflecting rising industry attention to agent runtime efficiency. A shift from competing on model capability to competing on the runtime and tooling layer could become an important trend for the next phase of AI infrastructure. Most of these products are at an early launch stage, and actual adoption and performance remain to be seen.