Run, Build, and Manage Your AI Agents
Grade: A — Score: 90/100
Agent Hub leverages advanced AI technology built on your CRM data, enabling agents to operate with contextual understanding to drive effective results across various business functions.
The platform simplifies workflows by allowing teams to manage multiple AI agents from a single interface, automating tasks such as campaign planning, prospecting, and customer support, thus freeing up valuable time for strategic activities.
With a focus on security and transparency, Agent Hub ensures that user data is protected and that businesses maintain control over their information, minimizing risks associated with AI deployment.
Starter Agent Access: No standalone Agent Hub fee; current HubSpot pages disagree on whether the full Agent Hub workspace is included with Starter
Professional / Enterprise Agent Hub: Included with eligible HubSpot Professional or Enterprise subscriptions; underlying HubSpot subscription pricing varies by product
HubSpot Credits: $0.010 per Credit; 1,000-Credit capacity packs cost $10, while current HubSpot pricing pages show $9 per 1,000 when paid annually
Consider switching to Salesforce Einstein: Salesforce offers a more extensive suite of CRM tools integrated with AI capabilities.
HubSpot's current public sources conflict on this point. The main Agent Hub product page says Agent Hub is available to Starter, Professional, and Enterprise customers, while the current August 21, 2026 Agent Hub help article and dedicated Agent Builder page list the full workspace for Professional and Enterprise. HubSpot's Product & Services Catalog separately lists several individual pre-built agents for qualifying Starter products, so Starter buyers should verify whether they receive the full Agent Hub workspace or only specific agents.
HubSpot lets you simulate an agent run without consuming Credits, then shows an estimated Credit cost before full execution. The estimate can differ from the actual run cost because agent complexity affects consumption. You can also set a monthly run limit for each agent so the account can cap how often that agent executes.
Yes, if the agent is given the required actions and permissions. Agent Builder can give a custom agent read access to HubSpot CRM records, write access to HubSpot CRM records, web browsing, and other HubSpot actions. A Super Admin or user with Agent Builder permission is required to create or customize agents, and additional permissions may be required for the actions the agent performs.
Yes. HubSpot's Run Agent workflow action can pass workflow data to an agent and use the generated output in later automation steps such as record updates or follow-up actions. The current July 22, 2026 documentation lists a limit of 500 Run Agent executions per day and requires HubSpot Credits.
HubSpot MCP Client currently documents support for Notion, Atlassian, Asana, Zapier, G2, Linear, Gong, and Amplitude. Depending on the MCP server, agents can query or update external data, create or manage tasks and issues, trigger automations, analyze sales calls, or query product analytics. HubSpot's MCP documentation lists Starter eligibility for several Hubs even though its Agent Hub and Agent Builder pages place the full management and custom-building experience on Professional and Enterprise, so Starter MCP access should be verified for the specific subscription.
Yes. HubSpot's August 21, 2026 Agent Hub documentation labels Agent Hub as beta, and its July 24, 2026 Agent Builder documentation labels custom agents as beta. Professional and Enterprise subscriptions are the strongest documented path for custom Agent Builder access, and HubSpot Credits are required when custom agents run.
Yes. HubSpot says Super Admins can turn off AI Model Training at the account level, after which eligible customer data is no longer used for future training of HubSpot's own AI models, and opting out does not remove access to AI features. The setting is on by default, previously used data cannot be removed from already trained models, and accounts with Sensitive Data enabled are automatically opted out and cannot opt in.
HubSpot positions custom agents for work that requires analyzing data, generating structured outputs, taking actions, or making decisions from multiple data sources, while standard automation and workflows are included with eligible Professional and Enterprise subscriptions without consuming HubSpot Credits by themselves. Running an agent inside a workflow does require Credits, so deterministic steps that do not need AI reasoning can be cheaper to keep as ordinary workflow actions. Agent Hub is more appropriate when the task benefits from CRM context, web research, unstructured inputs, or judgment across several sources.
HubSpot Agent Hub is tied to eligible HubSpot subscriptions and is designed around HubSpot CRM context, while Microsoft Copilot Studio is sold as a standalone tenant-wide agent platform and supports external channels such as websites, apps, and social platforms. Microsoft lists 25,000 Copilot Credits at $200 per pack per month billed annually, with pay-as-you-go also available; HubSpot instead bundles Agent Hub access with qualifying subscriptions and meters many agent outcomes through HubSpot Credits. Copilot Studio therefore offers a clearer standalone deployment path, while HubSpot Agent Hub has the advantage of native HubSpot customer data and workflows.