Last Updated: 22 August 2026

Executive Summary

HubSpot has spent two years turning a CRM into an AI platform, and in July 2026 it retired the Breeze brand entirely, replacing it with Agent Hub. The AI is genuinely good, largely because it runs on your own CRM data rather than a generic model. Its outcome-based pricing — you pay when an agent resolves a ticket, not when it tries — is one of the fairest models in business AI. The problem is everything around it. Seat tiers, mandatory onboarding fees, marketing contact bands and non-rolling credits stack into a bill that rarely resembles the advertised price.

Estimated reading time: 10 minutes

Key Takeaways

• HubSpot retired the “Breeze” AI brand in late July 2026. The same agents now sit under Agent Hub, alongside a new no-code Agent Builder in public beta for Professional and Enterprise customers.
• Outcome-based AI pricing is genuinely buyer-friendly: roughly $0.50 per resolved conversation, $1.00 per lead and $0.10 per answer, charged only on success.
• The free CRM is real and unlimited on users. Most AI agents, however, require Professional, where prices jump sharply.
• Mandatory onboarding fees of $1,500 to $7,000 apply to Professional and Enterprise, and credits do not roll over month to month.
• Agents that were free during the Breeze Studio beta started consuming credits in July 2026 — a quiet cost increase worth auditing.
• HubSpot’s business is growing fast, but its shares fell around 48% across 2026, and it reversed a customer data-sharing plan after backlash.

What Is HubSpot?

HubSpot is a customer platform: CRM at the centre, with marketing, sales, service, content, data and now revenue tooling built around it. Its long-standing appeal is that a small business can run its entire go-to-market operation in one place without stitching six products together.

Since 2024, the company has repositioned as an AI-first platform. That started with Breeze, launched at INBOUND 2024, which bundled an AI assistant, autonomous agents and data enrichment. In July 2026 HubSpot retired the Breeze name and reorganised the same capabilities under Agent Hub — announced on 23 July as a public beta for Professional and Enterprise customers, alongside Agent Builder, a no-code canvas for creating custom agents.

The strategic logic is straightforward. An AI agent is only as useful as the data behind it, and HubSpot already holds the customer records, email history, deals and tickets. Agents grounded in that data can answer questions and take actions a general assistant cannot.

The company reported first-quarter 2026 revenue of $881 million, up 23% year on year, with enterprise deals above $120,000 in annual recurring revenue growing 64%. It has bought capability aggressively too — Dashworks in 2025, XFunnel in early 2026 to anchor an answer-engine optimisation suite, and Warmly in June 2026 for visitor identification.

Why this matters for your business: HubSpot is no longer selling software you operate. It is selling agents that operate on your behalf, priced by outcome. That changes the buying question from “how many seats do we need?” to “how much work do we want done, and what is each completed task worth to us?”

Who Is It Best For?

HubSpot is a strong fit for:

• Small and mid-sized businesses wanting one platform. The core argument is unchanged and still good: marketing, sales and service in one system beats integrating three.
• Support teams with a real ticket volume. The Customer Agent reportedly resolves around 65% of conversations on average, and outcome pricing means you pay only for resolutions.
• Businesses with clean CRM data. Agents grounded in accurate records perform well. This is the single biggest predictor of whether HubSpot’s AI works for you.
• Teams that want AI without engineering. Agent Builder lets operations staff assemble custom agents in plain language rather than code.
• Companies starting from zero. The free CRM genuinely supports unlimited users and is a legitimate starting point, not a demo.

HubSpot is probably not the right choice for:

• Businesses with messy CRM data. Agents will automate the mess faster. Fix the data first.
• Very small teams needing agent capability, since most agents require Professional — a steep jump from Starter.
• Organisations that dislike mandatory fees. Onboarding at Professional and Enterprise is compulsory whether you want the help or not.
• Buyers who need predictable monthly costs. Between seats, contacts, credits and overages, this is a variable bill by design.
• Teams wanting deep customisation of AI behaviour, where reviewers note limited control over agent instructions.

Key Features

Agent Hub. The unified console for every AI agent in your account, organised by outcome rather than product — building demand, winning deals, delighting customers. Shows what is active, what each agent has done, and what has not been switched on yet.

Agent Builder. A no-code canvas, in public beta, for creating custom agents using plain language plus the data already in your CRM. HubSpot cites a customer that built an agent to parse school district calendars, cutting a 15 to 20 minute task to seconds and reclaiming an estimated 350 hours a year.

Customer Agent. The AI support agent, working across nine channels including chat, email, WhatsApp, Instagram, Telegram, SMS and voice in beta. Reportedly resolves around 65% of conversations on average, with strong implementations reaching 90%.

Prospecting Agent. Rebuilt to handle the sales lifecycle from identifying buying signals through to booking meetings and preparing pre-call briefings.

Data Agent and enrichment. Standard firmographic enrichment — revenue, industry, employee count, location — is now included with Core Seats rather than consuming credits, which is a genuine improvement over the earlier model.

Breeze Assistant. The in-app conversational helper for drafting, summarising records and searching. Available broadly, including on lower tiers.

Audit Cards. Transparency on what an agent actually did, showing which CRM properties were updated. A meaningful governance feature when agents are writing to your customer records.

HubSpot AEO. Answer-engine optimisation, built on the XFunnel acquisition, tracking how your brand appears when buyers ask AI assistants questions.

How the money actually works

SEATS ─────────► Core / Sales / Service / Revenue seats
   │              Tier applies to ALL paid seats
   ▼
CONTACTS ──────► Marketing Hub only, billed in 1,000 blocks
   │
   ▼
ONBOARDING ────► $1,500-$7,000 one-off. Mandatory at Pro+
   │
   ▼
CREDITS ───────► 500 / 3,000-5,000 / up to 10,000 per month
   │              Do NOT roll over. Overage ~$0.01 each
   ▼
OUTCOMES ──────► $0.50 resolution · $1.00 lead · $0.10 answer
   │              Charged on success only
   ▼
              YOUR ACTUAL BILL

[Illustration placeholder: HubSpot cost structure diagram]

Ease of Use

The product itself is easy, and that has always been HubSpot’s advantage over Salesforce. The interface is clean, non-technical staff learn it quickly, and the free tier means a business can start using the CRM the same afternoon.

The AI follows that pattern. Breeze Assistant needs no training, and Agent Builder is designed so that operations people rather than developers can assemble agents. Agent Hub adds something genuinely useful for governance: one console showing what every agent is doing, which matters as soon as you have more than one running.

Where HubSpot is not easy is commercial. Understanding what you will actually pay requires modelling seats, seat tiers, marketing contact bands, onboarding fees, monthly credit allowances and per-outcome charges together. Reviewers describe the pricing as a pyramid rather than a price list, and that is fair.

Future Signal Tip: Before enabling any agent, run it against your historical tickets or leads first. HubSpot lets you simulate before going live, which converts an unknown monthly cost into a number you can actually budget for.

Performance

The AI performs well when the underlying data is good, and poorly when it is not. That sounds obvious, but it is the most consistent finding across independent coverage: agents inherit the quality of your CRM. Inconsistent properties and duplicate records do not get fixed by adding AI, they get automated.

Where data is clean, results are real. The Customer Agent’s reported average resolution rate of around 65% is competitive with dedicated support AI tools, and the outcome-based model means poor performance costs you less rather than more. Agents run on current-generation models, upgraded during 2026.

The recognised weaknesses are about control rather than capability. Reviewers note there are limited custom instructions for shaping agent behaviour, which frustrates teams with specific tone or escalation requirements. Agent Hub and Agent Builder are also in public beta, so expect changes.

Integrations

This is HubSpot’s structural advantage. The App Marketplace runs to well over a thousand integrations, the APIs are mature and well documented, and most business software a small company uses has a HubSpot connector already.

The 2026 additions extend that into advertising and finance: Microsoft Ads, ChatGPT ads, and mapping fixes for QuickBooks and Xero. The newly launched Revenue Hub, described by HubSpot as the evolution of Commerce Hub, brings quote-to-cash and payments inside the platform, competing directly with Salesforce Revenue Cloud.

The trade-off is the familiar platform one. The deeper you integrate, the more expensive leaving becomes, and HubSpot’s commercial model rewards consolidation onto its own hubs rather than best-of-breed tools around it.

Pricing & Value

HubSpot’s pricing is the hardest part of this review, and the part most worth reading carefully. The advertised price is rarely the real one.

Four things compound on top. Seat tiers apply to everyone — if one person needs a Professional feature, every paid seat moves to Professional pricing. Marketing contacts are billed in blocks of 1,000 on Marketing Hub. Credits do not roll over, and overage runs at roughly $0.01 each, or about $9 to $10 per thousand. Onboarding is mandatory at Professional and above.

The genuinely good news is the AI pricing itself. Since April 2026, HubSpot charges by outcome: about $0.50 per resolved conversation, $1.00 per qualified lead and $0.10 per answer, billed only when the agent succeeds. Compared with per-seat AI licensing, that is a fairer deal, and it survived the July rebrand unchanged.

⚠️ The quiet July change. Agents that ran free during the Breeze Studio beta began consuming credits when Agent Hub launched. If you built agents during the beta, audit your credit consumption now rather than at renewal.

One worked example from independent analysis is instructive: a ten-agent support team on a headline “$900 a month” plan landed near $18,300 in the first year once seats, onboarding and AI usage were counted. What does your real number look like once you add onboarding and credits?

Strengths

• Agents run on your own CRM data, producing grounded output a general AI assistant cannot match.
• Outcome-based AI pricing charges only on success — among the fairest models in business AI.
• The free CRM is genuinely unlimited on users and a legitimate long-term starting point.
• Agent Builder makes custom agents accessible to operations staff without engineering support.
• Audit Cards and Agent Hub provide real governance over what agents are doing to your data.

Limitations

• Total cost is genuinely hard to predict, combining seats, contacts, credits, overages and fees.
• Onboarding fees of $1,500 to $7,000 are mandatory at Professional and Enterprise.
• Credits expire monthly, and all AI features draw on the same shared pool.
• Most agents require Professional, a steep jump from Starter pricing.
• Limited custom instructions for controlling agent behaviour and escalation.
• Agent Hub and Agent Builder are public beta, so behaviour and packaging may change.

Alternatives

Salesforce with Agentforce — the comparison at the top end. More configurable and more powerful at enterprise scale, and considerably harder and more expensive to run.

Zoho CRM — dramatically cheaper with respectable AI, and the sensible choice when budget matters more than ecosystem depth or polish.

Pipedrive or Close — better for sales-led teams that want a focused CRM rather than a marketing and service platform.

A dedicated support AI alongside a cheaper CRM — worth modelling seriously. Specialist tools often price support resolution below HubSpot’s effective cost once Professional seats are included.

The Future Signal

The Breeze-to-Agent-Hub rename in July 2026 looks cosmetic and is not. HubSpot has stopped organising its AI around a product brand and started organising it around outcomes — build demand, win deals, delight customers. That is a deliberate signal about what it now thinks it sells.

The pricing follows the same logic. Charging per resolved ticket rather than per seat is HubSpot betting that AI does measurable work, and accepting the revenue risk if it does not. That is a genuinely customer-aligned move and rare enough to note.

The tension is that everything around the AI still prices like traditional software: mandatory onboarding, tiered seats, contact bands, expiring credits. A business buying outcome-priced AI on top of a subscription structure designed for a different era gets both models’ complexity at once.

Two things are worth watching. HubSpot’s shares fell roughly 48% across 2026 despite 23% revenue growth, which tends to produce monetisation pressure. And the company reversed a customer data-sharing plan after backlash this year — a reminder that when a vendor holds your customer data, its strategy becomes your governance problem.

Final Verdict

Overall Rating: 8.0 / 10

HubSpot remains the most approachable serious customer platform available, and its AI is better than most because it is grounded in data you already own. Outcome-based pricing is a genuine credit to the company and should be the industry norm.

It loses points on commercial clarity rather than capability. Mandatory onboarding fees, expiring credits, tier-wide seat pricing and a steep Starter-to-Professional jump make the real cost difficult to forecast, and the July credit change to previously free beta agents was not well signposted.

Future Signal recommends HubSpot for small and mid-sized businesses wanting one platform, particularly support teams with real ticket volume. Model the full first-year cost, including onboarding and credits, before signing anything.

Before you buy — a short checklist

• Audit your CRM data quality first. Agents amplify whatever is already there
• Model year one fully: seats, contacts, onboarding, credits, overages
• Check which agents you need and whether they require Professional
• Remember tier pricing applies to every paid seat, not just the one that needs it
• Simulate agents against historical tickets before going live
• If you built agents in the Breeze Studio beta, audit their credit use now
• Compare against a cheaper CRM plus a specialist support AI
• Confirm current prices directly — HubSpot’s packaging changed twice this year

Frequently Asked Questions

Is Breeze still called Breeze? Mostly not. HubSpot retired the Breeze brand in late July 2026 and moved the agents under Agent Hub, launched 23 July as a public beta for Professional and Enterprise. Breeze Assistant remains as the in-app helper, so both names are currently in circulation.

How much does HubSpot’s AI actually cost? AI runs on HubSpot Credits, included monthly by tier and charged at roughly $0.01 each beyond that. Key agents use outcome pricing: about $0.50 per resolved conversation, $1.00 per lead and $0.10 per answer, billed only on success.

Is the free plan genuinely free? Yes. The free CRM supports unlimited users with contacts, companies and deals, with no time limit. Most AI agents, however, require Professional, so the free tier is a starting point rather than an AI evaluation.

Why is Professional so much more expensive than Starter? Because that is where automation, custom reporting and most agents live. The jump is large — Marketing Hub Professional starts around $800 a month against roughly $20 a seat on Starter — and it comes with a mandatory onboarding fee.

Do HubSpot credits roll over? No. Credits reset monthly and all AI features draw from the same pool, so heavy use of one agent reduces what is available for another. Budget for the peak month, not the average.

Can I control how HubSpot’s AI agents behave? Only partly. Agent Builder allows custom agents, and Agent Hub provides governance and Audit Cards, but reviewers consistently note limited custom instructions for shaping tone and escalation on the core agents.

Is HubSpot financially stable? Revenue is growing strongly — $881 million in Q1 2026, up 23% year on year — though the shares fell around 48% across the year. There is no viability concern, but sustained share pressure often precedes monetisation changes, so watch renewal terms.

Should I worry about my customer data? HubSpot proposed a customer data-sharing arrangement in 2026 and reversed it after customer objections. That is arguably reassuring on responsiveness, but it is a reminder to read data terms at renewal when a vendor holds your customer records.

Conclusion

HubSpot’s strength is that its AI is grounded in something real. Agents that read your actual deals, tickets and contact history produce work a generic assistant cannot, and Agent Hub finally gives businesses one place to see what those agents are doing. The outcome-based pricing genuinely deserves credit.

Its weakness is that the AI sits on top of a commercial structure built for a different era. Seat tiers that escalate across a whole team, mandatory onboarding fees, marketing contact bands and credits that vanish monthly combine into a bill few buyers forecast correctly the first time.

The practical approach is unglamorous but effective: clean your CRM data before enabling anything, simulate agents against historical work to get a real usage number, and model the full first year rather than the monthly headline. Do that and HubSpot is a strong, capable choice. Skip it and you will discover the pyramid the expensive way.