Quick answer
Agent Hub is HubSpot's home for AI agents — included with Professional and Enterprise subscriptions, with usage billed per action in HubSpot credits: roughly $0.50 per Customer Agent resolution, $1.00 per Prospecting Agent lead and $0.10 per Data Agent answer. It ships prebuilt agents for prospecting, customer service, data answers and AI-search visibility, plus a no-code Agent Builder for custom agents on your own CRM data.
HubSpot's AI story used to be a copilot in the corner of the screen. In 2026 it's a roster of agents that do actual work — resolve tickets, launch outreach, answer data questions — and Agent Hub is where they live.
This post covers what Agent Hub is, what the prebuilt agents actually do, how the credit-based pricing works, and the rollout pattern that keeps agents useful instead of expensive.
What is HubSpot Agent Hub?
Agent Hub is the home for AI agents inside HubSpot. Instead of bolting a chatbot onto your website, HubSpot builds agents directly on the customer platform: they read the same CRM records, deal pipelines, tickets and knowledge sources your team works in, and they write their results back to the same place.
That placement is the point. An agent that lives where your data lives doesn't need a sync, a middleware subscription or a separate login — and its work shows up in the reports you already run.
The prebuilt agents, and what they actually do
Prospecting Agent
Monitors buying signals across your CRM and launches personalized outreach — researching accounts, drafting sequences and keeping pipeline moving while your reps focus on live conversations.
Customer Agent
Resolves service inquiries 24/7, answering from your knowledge base and CRM context, and escalates to a human when it isn't sure. This is the agent with the clearest per-action economics: it either resolves a ticket or hands it over.
Data Agent
Answers questions from your CRM data, call recordings and documents on demand — "which open deals haven't been touched in 14 days?" — without anyone building a report first.
HubSpot AEO
Tracks how visible your brand is in ChatGPT, Gemini and Perplexity answers — the AI-search counterpart to rank tracking, increasingly important as buyers ask AI assistants instead of searching.
Agent Builder: custom agents without code
Where the prebuilt agents stop, Agent Builder starts: a no-code builder for custom agents on your own prompts, knowledge sources and CRM scope. The best builds are narrow and boring in the best way — a renewals agent that flags at-risk accounts, an enrichment agent that fills missing firmographics, an internal FAQ agent for the sales team.
How Agent Hub pricing works
Agent Hub is included with Professional and Enterprise subscriptions. Usage is billed per action in HubSpot credits, at published rates of roughly:
| Agent action | Published rate |
|---|---|
| Customer Agent — resolved conversation | ~$0.50 per resolution |
| Prospecting Agent — qualified lead | ~$1.00 per lead |
| Data Agent — answer | ~$0.10 per answer |
Two things matter more than the rates themselves. First, you pay for outcomes — a resolution, a lead, an answer — not for tokens, which makes the spend legible to a CFO. Second, credit allocations don't roll over, so the real discipline is capacity planning: estimate volume per agent, set alerts, and review burn monthly.
Does it work? What HubSpot reports
HubSpot's own numbers: customers with agents in place close 77% more support tickets monthly and create 65% more sales leads. Treat vendor-reported figures as directional rather than gospel — but the pattern matches what we see in real portals: agents are good at volume work, and the wins come from letting humans handle the exceptions instead of the queue.
Three readiness checks before you switch anything on
1. Data quality. Agents answer from your CRM. Duplicates, stale records and undocumented properties turn directly into wrong answers — at scale, politely, 24/7. Clean the data the agents will read first.
2. Guardrails. Decide before launch what agents may do alone, what needs an approval step, and when they must escalate to a human. Brand-safe response boundaries are configuration, not hope.
3. Credit strategy. Per-action pricing is predictable only if someone owns the forecast. Set a monthly credit budget per agent and review it like ad spend.
A rollout that works: pilot, measure, scale
The failure mode is switching everything on for everyone. The rollout that works is the same one that works for a new Hub: pick one team, run one agent on real work, track outcomes and credit burn for a few weeks, then scale what earned it.
If you want the guided version of exactly that — readiness check, agent configuration, guardrails, credit strategy and team training — that's our Agent Hub onboarding.
Frequently asked questions
Is Agent Hub included in my HubSpot subscription?
Agent Hub is included for Professional and Enterprise customers. You don't buy the agents themselves — you pay for what they do, per action, in HubSpot credits.
Which prebuilt agents does HubSpot ship?
Prospecting Agent (buying signals and outreach), Customer Agent (24/7 service resolution), Data Agent (instant answers from CRM data, calls and documents) and HubSpot AEO (brand visibility in ChatGPT, Gemini and Perplexity answers) — plus Agent Builder for custom no-code agents.
What do HubSpot AI agents cost to run?
Published per-action rates are roughly $0.50 per Customer Agent resolution, $1.00 per Prospecting Agent lead and $0.10 per Data Agent answer, billed in HubSpot credits. Allocations don't roll over, so plan for capacity you'll actually use.
Is our portal ready for AI agents?
Check three things: the data agents will answer from (duplicates and stale records become wrong answers), the guardrails (approvals, escalation, response boundaries) and a credit budget with an owner. If any of the three is missing, fix that before switching agents on.