On July 23, 2026, HubSpot rolled out Agent Hub and Agent Builder in public beta for every Professional and Enterprise customer — not as another single-purpose bot, but as the control tower for every AI agent already living inside a company’s CRM. The timing is not a coincidence. Three weeks earlier, on June 30, HubSpot agreed to acquire Warmly, the AI-native go-to-market platform known for identifying more than half of a website’s anonymous visitors by name. Put the two announcements side by side and a strategy comes into focus: HubSpot is no longer shipping isolated AI features. It is building a coordinated workforce of agents that all read from, and write back to, the same customer record.
Key Takeaways
- Agent Hub is a single dashboard for monitoring every HubSpot AI agent — prebuilt and custom — with live status, performance data, and one-click activation.
- Agent Builder is a no-code, natural-language canvas for building custom agents that run on a company’s existing Smart CRM data, with no separate field mapping or integration setup.
- Both tools are in public beta as of July 23, 2026, included for Professional and Enterprise subscribers; custom agent actions draw on HubSpot Credits.
- The June 30, 2026 acquisition of Warmly adds an Inbound Agent, a TAM Agent, and person-level visitor identification to HubSpot’s agent portfolio.
- Early customer Ignite Reading reports reclaiming roughly 350 hours a year with a single custom agent built in the new tools.
- The core problem HubSpot says it is solving is agent fragmentation — multiple bots operating on different, incomplete pictures of the same customer.
Most CRM buyers who added an AI agent in the last two years did it one tool at a time: a chatbot here, a lead-scoring model there, a meeting scheduler bolted on somewhere else. Each one worked from its own slice of data. HubSpot’s Chief Product and Technology Officer Duncan Lennox named that problem directly at launch: “The problem isn’t managing a single agent in isolation. It’s that once you have multiple agents, they become fragmented, all working from different pictures of the customer, or even worse, no picture at all.” Agent Hub and Agent Builder are HubSpot’s answer to that fragmentation, and they matter to CRM buyers regardless of vendor, because every CRM platform is racing toward the same multi-agent future.
What Agent Hub and Agent Builder Actually Do
Agent Hub is the management layer. It surfaces every featured HubSpot agent — prebuilt and custom — in one dashboard, showing live status and recent results, and organizing them not by department but by commercial outcome: building demand, winning deals, delighting customers, and scaling growth. Any agent sitting dormant can be switched on with one click, and Agent Hub also serves as the front door to HubSpot’s Agent Marketplace and to Agent Builder itself.
Agent Builder is the construction layer. It replaces the old pattern of stitching together workflows, custom code, and third-party automation tools with a single canvas where teams describe what they want an agent to do in plain language, using HubSpot’s Breeze Assistant to assemble the logic. Triggers can be a scheduled time, a contact property update, an inbound webhook, or an event from a connected third-party tool. Because the agent runs natively inside Smart CRM, it already has access to deal history, contact records, call transcripts, and buying signals — there is no separate data pipeline to build or field mapping to maintain. HubSpot has also built in guardrails that require human approval before an agent takes an autonomous action, which matters for any team nervous about a bot updating records or contacting customers unsupervised.
The featured early example is Ignite Reading, a virtual literacy tutoring program operating across 25 states. The organization built a custom agent that parses school district academic calendars — a task that used to take a staff member 15 to 20 minutes per district — and now completes it in seconds, an estimated savings of more than 350 hours a year. It is a mundane, unglamorous use case, and that is exactly why it is a useful proof point: the value isn’t a flashy chatbot demo, it’s hours of manual, repetitive data entry disappearing from someone’s week.
Why the Warmly Acquisition Is Part of the Same Story
HubSpot’s June 30 agreement to acquire Warmly is easy to read as an unrelated tuck-in deal, but it plugs directly into the fragmentation problem Agent Hub is meant to solve. Warmly’s technology identifies more than half of a website’s anonymous visitors at the individual level — turning an unnamed hit in the analytics dashboard into an actual name and company. It ships with two agents of its own: an Inbound Agent that engages identified visitors in chat and logs the transcript straight into Smart CRM, and a TAM Agent that proactively identifies and engages likely buyers in a target account list before they ever land on the website. Strung together, the logic is a closed loop: the TAM Agent fills the top of the pipeline, visitor identification tells sales who just showed up and how warm they are, and the Inbound Agent converts the visit into a conversation — all of it landing in the same CRM record that Agent Hub is now built to supervise. HubSpot has said existing Warmly contracts, pricing, and product experiences remain unchanged for now, with deeper integration planned over time.
For a CRM buyer, the acquisition is a signal worth paying attention to independent of Agent Hub: HubSpot is willing to buy point-solution AI capability rather than build every piece from scratch, and it is prioritizing the go-to-market funnel — top-of-funnel targeting, mid-funnel conversion, and CRM record-keeping — as one connected system rather than three separate purchases.
Why This Matters for CRM and ERP Buyers Generally
Even organizations with no plans to touch HubSpot should pay attention, because the underlying pattern is now showing up across the CRM and ERP landscape: platforms are shifting from selling individual AI features to selling agent orchestration. A company evaluating any CRM or ERP AI roadmap should be asking the vendor a version of the same question HubSpot is answering here: when we have five or ten agents running, who tells us what they’re doing, whether they’re working from the same data, and how we shut one off without breaking the others?
That question has real operational weight. An agent that scores leads using stale contact data, while a second agent is simultaneously updating that same contact record based on a different signal, produces exactly the kind of confusion Lennox described — bots stepping on each other because none of them share a single source of truth. Centralized agent management is what prevents that from becoming an invisible, hard-to-diagnose data quality problem six months into a rollout.
Practical Use Cases Beyond the Launch Example
- Manufacturing and distribution: a custom agent that watches for repeated stockout patterns on high-velocity SKUs and automatically drafts a reorder recommendation for a sales rep to approve, rather than waiting for a manual inventory review.
- Professional services: an agent triggered on contract renewal dates that pulls usage and support-ticket history into a renewal risk summary before the account manager’s first call.
- Education and nonprofits: following Ignite Reading’s example, agents that ingest structured but inconsistent external data — academic calendars, grant reporting formats, compliance deadlines — and convert it into CRM-native records without a developer.
- Home builders and construction: a lead-routing agent that combines Warmly-style visitor identification with a buyer’s browsing behavior on floor plan pages to flag serious prospects to a sales rep same-day instead of at end-of-week reporting.
- Healthcare-adjacent services businesses: an intake agent that pre-populates a new patient or client record from a form submission and books a scheduling agent for confirmation, with a human approval step before any outbound message goes out.
Benefits and Challenges
The benefit case is straightforward: less time spent on repetitive data entry, faster response to inbound signals, and a single place to see whether the AI investment a company already made is actually running or sitting dormant. The natural-language, no-code construction model also lowers the bar for who inside an organization can build an agent — it does not have to be a developer or a HubSpot administrator with workflow expertise.
The challenges are just as real. Public beta software changes quickly, and teams that build heavily customized agents now should expect behavior, pricing, or credit consumption to shift before general availability. The credit-based pricing model for custom agent actions means usage needs to be monitored the same way a company would monitor API or automation consumption elsewhere — an enthusiastic rollout of five or six custom agents without a usage budget can produce a bill surprise. And guardrails requiring human approval are a safety feature, but they are only as good as the team’s discipline in actually reviewing what an agent proposes rather than rubber-stamping every suggestion.
| Capability | Agent Hub | Agent Builder |
|---|---|---|
| Primary function | Monitor and manage all agents in one place | Create and customize new agents |
| Who uses it | RevOps, marketing/sales/service leaders | Anyone building a workflow, no code required |
| Data source | Aggregated status across all connected agents | Native Smart CRM data — deals, contacts, calls, signals |
| Trigger control | Activate/deactivate existing agents | Scheduled times, property changes, webhooks, third-party events |
| Availability | Public beta, Professional & Enterprise | Public beta, Professional & Enterprise |
Implementation Best Practices and Common Mistakes
Start with one well-defined, currently manual process — the way Ignite Reading targeted a specific calendar-parsing task — rather than trying to automate an entire department at once. A narrow, measurable win builds internal trust in agent output before anyone hands over higher-stakes decisions. Audit existing CRM data quality before turning an agent loose on it; an agent built on top of duplicate contacts or stale deal stages will simply automate the mess faster. Assign clear ownership for every agent in Agent Hub — someone needs to be accountable for reviewing performance and turning an underperforming agent off, rather than letting dormant or misbehaving agents accumulate unnoticed. And treat the approval-guardrail setting as a policy decision, not a default to click past: decide deliberately which actions are safe to fully automate and which should always route to a human first.
The most common mistake we expect to see is treating Agent Builder as a workflow replacement without rethinking the workflow itself. Companies that just recreate their existing HubSpot workflows inside the new agent canvas capture little of the value; the bigger win comes from identifying tasks that were previously too unstructured or judgment-heavy for a rules-based workflow but are now tractable for a natural-language agent working off full CRM context.
CRM Experts Online’s Perspective
We’ve implemented enough HubSpot portals to know that the gap between “we turned on an AI feature” and “we changed how the team actually works” is almost always a data and governance gap, not a technology gap. Agent Hub is a genuinely useful answer to a problem we see constantly in client portals: three or four AI-adjacent tools enabled over time by different people, none of them coordinated, and nobody quite sure which ones are still active or whether they’re stepping on each other. Having one screen that shows every agent’s status and lets a RevOps lead switch one off without archaeology through settings menus is a real operational improvement, not just a UI refresh.
Our advice to clients evaluating this rollout is the same advice we’d give for any new automation layer: pressure-test the data foundation first. An agent built in Agent Builder is only as reliable as the deal stages, contact properties, and call transcripts it’s reading from, and we routinely find those in worse shape than clients expect going into a project like this. We also recommend piloting with the approval guardrails turned all the way up before loosening them — earn the trust in agent output before removing the human checkpoint, not the other way around. If your team is looking at Agent Hub, Agent Builder, or the incoming Warmly capabilities and wants a second set of eyes on what to automate first and how to structure the guardrails, that’s exactly the kind of implementation planning we do daily.
FAQ
Is Agent Hub available on every HubSpot tier? No. It’s included in the public beta for Professional and Enterprise editions; lower tiers are not currently eligible.
Does using Agent Builder require a developer or HubSpot admin certification? No. It’s designed as a no-code, natural-language canvas, though teams with workflow and data hygiene experience will get better results faster.
How is custom agent usage billed? Custom agents built in Agent Builder consume HubSpot Credits when they perform configured actions. Credits are included at the Professional and Enterprise tiers, with additional credits purchasable.
Can an agent take action without human review? HubSpot has built in guardrails requiring approval before autonomous actions, though the level of autonomy for any given agent is a configuration decision each team makes.
What happens to Warmly customers after the acquisition? HubSpot has said existing Warmly contracts, pricing, account teams, and product experiences remain unchanged for now, with deeper platform integration planned over time.
Is this the same thing as HubSpot’s earlier Breeze Agents? Breeze Agents (prospecting, customer, content) are the prebuilt agents that now live inside Agent Hub alongside any custom agents a team builds; Agent Hub is the new umbrella for managing all of them together.
Should we wait for general availability before building anything? Not necessarily — starting with a low-risk, narrowly scoped pilot during the beta is a reasonable way to learn the tool, as long as you’re prepared for behavior or pricing details to evolve before GA.
Does this replace the need for a CRM administrator or implementation partner? No. It lowers the technical bar for building an individual agent, but data architecture, governance, and prioritization decisions still benefit from experienced oversight.
Conclusion
Agent Hub and Agent Builder are less about a single new feature and more about HubSpot admitting what every CRM buyer running multiple AI tools already suspected: uncoordinated agents create as much confusion as they remove. Combined with the Warmly acquisition, HubSpot is betting that the winning CRM platform of the next few years isn’t the one with the most individual AI features, but the one where every agent shares the same picture of the customer. If your team is running HubSpot, or evaluating a multi-agent AI strategy on any CRM platform, now is a good time to audit what’s already automated, what’s still fragmented, and where a properly governed agent could reclaim real hours — not just headlines. Schedule a consultation with CRM Experts Online to map out an agent rollout that starts with your data foundation, not just the demo.
Further Reading
- HubSpot: Meet Agent Hub and Agent Builder
- Martech Notes: HubSpot launches Agent Hub and Agent Builder in public beta
- MarTech: HubSpot’s Warmly deal points to the next generation of CRM
- Warmly: Warmly Is Joining HubSpot
- CX Today: HubSpot Acquires Warmly to Fix CRM’s Biggest Blind Spot
CRM & ERP Enterprise Technology Expert and Entrepreneurial Executive with 20+ years of leading CRM, ERP, Customer Experience, and Block-chain initiatives and projects across internal and customer facing technologies. Proven success in closing large deals in Pre Sales customer facing engagements and deploying enterprise wide CRM & Customer Experience solutions internationally and domestically.