What it is
A long-horizon agent is an AI agent built to pursue one goal over days, weeks or even months instead of a single chat session. In Agentforce, Salesforce announced long-horizon agents at Dreamforce 2026: you assign a goal, the agent builds a multi-step plan, executes it, and checks in for approval based on guardrails you define.
Why it matters
Most agents forget everything when a session ends. Goals like re-engaging at-risk deals take weeks, so the agent needs memory, a plan that survives failures, and a way for people to adjust direction along the way.
Key components
- Memory across sessions
- Durable execution that course-corrects when a step fails
- Dynamic steering in plain language
- Approval check-ins based on guardrails
How it connects
Salesforce runs these agents on a new long-horizon runtime under Agentforce. Hunter, the outbound sales agent, is the first to use it (pilot at announcement, GA targeted for November 2026).
Good to know
Settle approval rules, action design and data quality before turning one on. An agent that works for weeks repeats any bad data or loose action it is given.
Frequently asked questions
What is a long-horizon agent in Agentforce?
It is an Agentforce agent that pursues a single goal over days, weeks or months, using memory across sessions, durable execution and plain-language steering, and checking in for approval based on your guardrails.
Related terms
Agent Memory
The systems that let an AI agent retain context across calls, sessions, users, or teams — turning a stateless model into something with continuity. Encompasses short-term working memory, long-term episodic memory, and shared organizational memory.
Agentforce
Salesforce's AI agent platform that enables businesses to build, customize, and deploy autonomous AI agents across sales, service, marketing, and commerce.
AI Agent
An autonomous AI system that can perceive its environment, make decisions, and take actions to achieve specific goals - without constant human direction.
Guardrails
The boundaries you set on AI agents to control what they can discuss, what actions they can take, and when they must escalate to a human.
Human-in-the-Loop (HITL)
A design pattern where AI systems require human approval or intervention at critical decision points before taking action.
Long-Horizon Agent in practice
Articles from the Funnelists Team that put this term to work.
