Sign InOpen Brain
AI EngineerVideoSource Linked

Scaling to Long Horizons — Ross Taylor & Chengxi Taylor, General Reasoning

Long-running agents strain context, reward design, and GPU scheduling at once. Compaction and value models can help, but each introduces another optimization trade-off.

AI Engineer · Jul 31, 2026
Open Source Open MarkdownOpen JSON
Source Summary

Long-horizon RL faces gradient variance that grows with trajectory length, sparse terminal rewards, credit assignment, and variable episode lengths. A **1 million-token context window** is still small relative to the horizons the speakers target.

Practical Implication

Plan agent state beyond the prompt: use compaction, file-based scratch space, trajectory search, or archives as the task permits. For training, critics can provide earlier signals, while pipeline RL trades fresher policy data for better GPU use.

Agent-Ready Context
Long-horizon RL faces gradient variance that grows with trajectory length, sparse terminal rewards, credit assignment, and variable episode lengths. A **1 million-token context window** is still small relative to the horizons the speakers target.

Plan agent state beyond the prompt: use compaction, file-based scratch space, trajectory search, or archives as the task permits. For training, critics can provide earlier signals, while pipeline RL trades fresher policy data for better GPU use.

The team reports pipeline training tolerating about **eight off-policy steps**, but week-long inference can exceed that window. Bootstrapping with a value model keeps hardware occupied at the cost of bias, and its platform’s **350 environments** do not remove that trade-off.
Connected Context · Feed7 Judgment

This establishes long-horizon agency as a state-management and training-systems problem that a larger prompt alone does not solve. It deepens prior harness guidance by requiring durable state outside context and exposes a deployment gap: pipeline RL may tolerate limited policy staleness, while week-long trajectories can exceed it, leaving a trade-off among hardware use, fresh data, critic bias, and sparse rewards.

Data and Environment Curation for Post-Training LLMs — Mahesh Sathiamoorthy, Bespoke LabsBespoke’s SFT-first guidance becomes more consequential here because long-horizon RL adds sparse rewards, growing gradient variance, policy staleness, and infrastructure costs beyond ordinary curation.Do AI Agents Know When a Task Is Simple? Toward Complexity-Aware Reasoning and ExecutionE3’s minimum-viable-path strategy can avoid unnecessarily long trajectories, whereas this signal addresses the durable state and training problems that remain when the task genuinely requires a long horizon.AI tools for Forward Deployed Engineering — Vasuman Moza, Varick AgentsVarick’s explicit mapping of ownership, exceptions, and handoffs provides the process structure that long-running agents would need to preserve through compaction, archives, or file-based state.Let's integrate AI Agents in Event-Sourced Systems — Divakar Kumar, FlyersSoftThe event-sourced design offers a concrete external-state pattern for bounded context and resumable decisions, complementing this signal’s warning that prompt context cannot hold the full operational horizon.
Context Map
agentcodingdata#harness-engineering#agent-reliability#context-engineering
Uncertainty
The team reports pipeline training tolerating about **eight off-policy steps**, but week-long inference can exceed that window. Bootstrapping with a value model keeps hardware occupied at the cost of bias, and its platform’s **350 environments** do not remove that trade-off.