About agents

Prev Next

Who can use this: Any team member

Available on: Any Gong plan

Ideal for: Sales manager, RevOps

Agents let team members turn recurring AI tasks into reusable workflows. Gong has two types of agents: on-demand agents and automated agents.

Many team members rely on Gong Assistant to analyze conversations, summarize deals, draft customer communications, or identify risks and opportunities. These tasks often require team members to enter similar prompts repeatedly. While the underlying goal stays the same, they must continually rewrite instructions, refine prompts, and explain the desired output.

Agents solve this problem by allowing team members to save instructions and reuse them whenever the same task needs to be performed again.

On-demand agents

On-demand agents are personal assets that team members create and manage for their own use. Any team member can create an on-demand agent, but each agent is personal and only available to the team member who created it. Agents are managed in Assets > Agents.

On-demand agents can be created through the Gong Assistant, the agent form, or a panel where Gong AI generates the agent for you to review.

Team members run on-demand agents from Gong Assistant to generate outputs based on their instructions and the available Gong context.

Automated agents

Required permissions: Manage automations

Automated agents run automatically based on a trigger you define. When the trigger fires, the agent carries out its actions. Automated agents are available to all team members in your company.

The trigger is the event that starts the agent. Each automated agent uses one of the following triggers:

  • A CRM record change

  • A date condition

  • A smart tracker detecting a signal on a call

You can add conditions to narrow the trigger's scope, for example limiting the agent to opportunities at a specific stage or accounts of a certain type.

When the trigger fires, the agent runs the actions you configured, in sequence. An agent can have one or more actions. Available actions include:

  • Generate an AI analysis

  • Send an internal email

  • Post a Slack message

  • Add a contact to a flow

  • Send a brief

  • Start a play

Not all actions are available for every trigger type. Every run is recorded in the agent's run history.

How agents support recurring work

Agents are designed for work that team members perform repeatedly.

For example, a sales representative may regularly ask Gong Assistant to draft follow-up emails after customer calls. A manager may repeatedly analyze conversations for coaching opportunities. A customer success manager may frequently review customer interactions to identify risks before a renewal.

Agents allow team members to save these instructions and reuse them whenever they are needed. By applying the same instructions each time they run, agents help produce consistent outputs while reducing the effort required to complete recurring tasks.

How agents work with Gong data

Agents run within Gong and use the context available at the time they run.

When an agent is launched from a call, the agent can use information from that conversation. When an agent is run from an account or opportunity, the agent can use the context available for that account or opportunity. This allows agents to produce results that are grounded in the revenue data already stored in Gong.

Agents access data based on the Gong permissions of the team member running them. They can only access data that team member is authorized to see.

Creating reusable expertise

An agent can be as simple as a call summarization workflow or as detailed as a structured analysis that follows a specific methodology.

For example, an agent can be instructed to identify customer concerns, highlight risks, prioritize next steps, and present the results in a consistent format. Each time the agent runs, it follows the same instructions while adapting the output to the available context.

This makes agents useful for standardizing how recurring work is performed, while still allowing the output to reflect the specific customer, deal, or conversation being analyzed.

Agent capabilities and limitations

Agents have the following characteristics:

  • Agents generate text-based outputs based on their instructions and the information available when they run.

  • On-demand agents are personal assets and are only available to the team member who created them. Automated agents are available to all team members in your company.

  • Team members can view their own agent interactions in their Gong Assistant session history. Agent activity isn't available in a centralized audit log.

  • On-demand agents can have files attached to provide context when they run.

  • On-demand agents don't support actions, triggers, or MCP integrations.

  • Running on-demand agents doesn't consume AI Credits. Running automated agents consumes AI Credits.