Updates blog — Thursday, May 21, 2026

HubSpot Updates: May 21, 2026

Twelve updates landed Thursday, and two threads run the strongest. The first is the agentic automation push picking up speed: a new "Run Agent" workflow action drops Breeze agents directly into the Workflows app, on the heels of last week's Agentic Automation Builder beta. The second is consolidation. Centralized data model management, Context Home, and the products work inside Breeze Assistant all reduce the number of places teams have to look to manage how the platform thinks. The release also includes custom-coded modules for quotes going live, a PDF exports update that is exactly what it sounds like, and a sandbox parity pair that closes long-standing gaps for Enterprise admins.

If you watched the show today, you already heard the conversation around six of these. If you did not: Watch the recap to hear our thoughts on these updates and more.

Platform Updates Detailed

Run Agent Workflows Action

The Run Agent workflow action lets you trigger AI agents directly inside the Workflows app, alongside CRM data and other workflow logic. Until now, automating an agent meant configuring a trigger inside Breeze Studio, which was powerful but constrained to the Breeze surface. This action brings the agent into the same place where the rest of automation lives. This is in public beta across Marketing, Sales, Service, Data Hub, and Smart CRM Professional and Enterprise tiers.

Why it matters: This is the connective tissue piece between the agentic future HubSpot has been telegraphing and the workflow surface most teams already trust. The Agentic Automation Builder that landed in private beta last week is the long arc, the canvas-first builder where agents are first-class steps. The Run Agent action is the short arc: if you are already inside the linear workflow builder, you can now slot an agent into the flow without learning a new tool. Worth flagging that this is where the structured-versus-unstructured trade-off shifts. Workflows have always been deterministic. Agents are not. The combination changes the kind of automation you can ship, and changes how teams need to think about testing it.

For more details: Run Agent Workflows Action public beta.

Centralized Data Model Management

A unified table and graph view for managing the entire data model, including objects, properties, associations, and conditional logic, from one place. This replaces a workflow where data-model management was spread across multiple settings pages with no consistent picture of how the pieces connect. The new view includes a table, a graph, and an expansion mode in the upper left. This is in private beta across all hubs and tiers.

Why it matters: This is one of the most requested admin updates in HubSpot, and the team has been heads-down on it long enough that some of the static data-model reference tools admins built outside HubSpot are about to become unnecessary. The honest take: a single place to see and manage the data model changes what an admin's day looks like. Auditing complex configurations, making a quick edit to a property group, or figuring out why an object is or is not active becomes a single-screen task instead of a five-tab one. The bigger thing this signals is that the platform is recognizing the data model as a first-class surface, not a settings detour.

For more details: Centralized data model management private beta.

Context Home

Context Home is the new centralized location for the context that powers HubSpot AI, replacing the previous AI Data Sources page under AI Settings. Customers can now view and manage all the context HubSpot holds on their business, including business information, customer context, and team and process context, in one place. This is live across all hubs and tiers.

Why it matters: Brand identity, brand overview, AI data settings, and knowledge sources have been scattered across the portal for as long as anyone has been using them. Consolidating them under one home is the same architectural move centralized data model management is making for objects and properties: take a sprawl and give it a single page. The reason it matters specifically here is that the context layer is what every Breeze experience reads from. If teams cannot find or maintain it, the AI gets worse. If you have not visited Context Home yet, it is worth a walkthrough this week.

For more details: Context Home live.

Custom-Coded Modules for Quotes Now Live

Custom-coded modules for quotes moves from public beta to live. Developers, admins, and HubSpot solutions partners can now build custom, interactive, data-rich elements that embed directly into quote templates or individual quotes. Modules can access CRM data, pull from external sources, capture inputs and push them back to the CRM, and present interfaces to quote creators while they build. This is available for Commerce Hub Professional and Enterprise.

Why it matters: The capability has been covered, but live status is the news. The strategic piece is the CMS-CPQ collision. CMS and CPQ have traditionally been completely separate sides of the business, and customizing quotes the way teams actually want has always required CMS architecture. By landing these modules on CMS, HubSpot is treating "different data that needs to come together" as the design principle. AI coding tools have also lowered the barrier to entry for custom modules considerably, but the underlying caveat still applies: custom modules are only as good as the data model behind them. CPQ implementations get burned when the build ships before the data is ready.

For more details: Custom-coded modules for quotes now live.

Dashboard PDF Exports Look Like Dashboards

Dashboard PDF exports now retain the full dashboard layout, including charts, tables, titles, dates, and filter context, as displayed on screen. Scheduled report exports sent via email use the same improved format. This is in public beta across all hubs and tiers.

Why it matters: Exported dashboards are the version of HubSpot data most leaders actually see. The formatting gap between the in-app dashboard and the PDF version has been a quiet drag on how much teams trust the platform as a reporting surface. This is the kind of update where the name does the work: it now behaves the way most people assumed it always did. Worth flagging if you run scheduled reports for executives or clients.

For more details: Dashboard PDF exports look like dashboards public beta.

Content Recommendations for Emails

Content recommendations analyzes past email sends and peer benchmarks to surface tactical, ready-to-implement suggestions for improving email performance. Each recommendation comes with a sourced explanation of why the change is valuable. This is in private beta for Marketing Hub Professional and Enterprise.

Why it matters: Email best practices are hard to keep alive inside an organization. They get lost when someone leaves, or buried in a document nobody opens. Surfacing them as in-product recommendations, sourced from your own send history and peer benchmarks, is a different shape than a static best-practice doc. The honest take: this is the kind of feature whose value depends entirely on the quality of the recommendations. If they are obvious, teams ignore them. If they are specific and actionable, they become a real learning surface. Worth requesting if you have a marketer who would benefit from a steady drip of suggestions tied to actual performance data.

For more details: Content recommendations for emails private beta.

Create, Edit, and Analyze Products in Breeze Assistant

Breeze Assistant can now analyze, create, and edit products directly from the sidebar, including more complex pricing mechanisms like tiered pricing. Instead of clicking through records to understand how a product drives or impacts revenue, you can ask Breeze conversationally and have it return the analysis or make the change. This is live across all hubs and tiers.

Why it matters: Products are at the center of every deal and quote, but configuring them correctly and understanding the downstream impact of a change has always required either a deep familiarity with the product library or a click-by-click investigation. Moving that work into Breeze removes the friction without removing the structure: the records still live where they live, the pricing still behaves the way it behaves, but the path to understanding or changing them is conversational. This is also a quiet signal that Breeze is being given write access to more revenue-adjacent objects, not just read access. Worth watching the cadence of similar updates.

For more details: Create, edit, and analyze products in Breeze Assistant live.

Prioritize Invoices with Breeze

Breeze can now prioritize open invoices by revenue impact and send personalized collection emails to accelerate accounts receivable. The aim is to focus AR work on the invoices that matter most, reduce manual work, and improve cash flow. This is in private beta. The tier list reads as all hubs and tiers, Commerce Hub Professional, and Commerce Hub Enterprise, which is an unusual combination and likely reflects that some specific invoicing capabilities are starting to require Commerce Hub even though invoices themselves do not.

Why it matters: Collections work is a real pain that no adjacent system is solving. QuickBooks and most ERPs hand the team a screen of open balances and leave them to do the discovery, the prioritization, and the customer outreach. None of that work happens near the customer data, which means teams jump between systems or call people who have context. Bringing prioritization, the personalized message, and the customer record into one place is where Breeze actually earns the AR use case. When you can communicate better, people pay, and they pay faster.

For more details: Prioritize invoices with Breeze private beta.

HubSpot Capital Financing for Commerce Hub Payments Customers

Commerce Hub customers using HubSpot payments or Stripe payment processing in the US and UK can now access capital financing directly within HubSpot, based on their payment history. This is in public beta. Because financing is underwritten on processing history, it is an org-level offer rather than a feature to enroll in.

Why it matters: This is a business update more than a software update, and that is the interesting part. Stripe Capital underwrites businesses based on processing history, and HubSpot is so deeply integrated with Stripe that it can surface that financing offer right inside the platform. For SMBs on Commerce Hub, the option to pull capital from inside the tool you already use, based on data the platform already has, is a meaningful shortcut compared to a separate application. It is also one of the levers HubSpot is pulling to make Commerce Hub stickier. The careful piece: how this lands with leadership matters. Capital is not always a yes, and how it gets pitched internally is the difference between a tool and a flag.

For more details: HubSpot Capital financing for Commerce Hub payments customers public beta.

Manage Stage Calculated Properties by Pipeline

You can now control which pipelines and pipeline stages generate stage calculated properties (SCPs). By default, SCPs are disabled for new custom pipelines and stages but can be enabled anytime. A new usage tracker monitors SCP count against the 4,500 property limit per object type. This is in public beta across Professional and Enterprise tiers for all hubs.

Why it matters: SCPs being created automatically for every pipeline and stage was a clean way for portals to silently rack up hundreds or thousands of properties they did not need. Giving admins control over which pipelines actually generate them is the right design. The usage tracker is the unsung piece: knowing where you stand against the property limit is the kind of visibility that has been missing for portals that scaled past a few pipelines. This is overdue, and worth a quick audit of your own portal to see which pipelines could now have SCPs turned off.

For more details: Manage stage calculated properties by pipeline public beta.

Playbooks Now Supported in Sandboxes

When you create a new sandbox, playbooks from your production portal are now automatically copied over, including their content and recommendation settings. This is in private beta across Enterprise tiers (Commerce, Content, Marketing, Data, Sales, Service, and Smart CRM).

Why it matters: Playbook configuration changes are a real thing teams want to test before pushing to production, and until now sandboxes did not carry playbooks across. Closing that gap is a quiet but meaningful sandbox parity improvement. The broader pattern, taken together with the multi-brand assets update below, is that the sandbox roadmap is getting filled in piece by piece. If you manage a complex Enterprise portal, the gap between sandbox and production is closing.

For more details: Playbooks now supported in sandboxes private beta.

Sandbox Support for Multi-Brand Supported Assets

When you create a sandbox, HubSpot now copies assets assigned to all brands, not just the default. Previously, workflows, emails, and forms assigned to non-default brands were missing from the sandbox, which made it difficult to test changes that affect the full production environment. This is in private beta across Enterprise tiers (Commerce, Content, Marketing, Data, Sales, Service, and Smart CRM).

Why it matters: If you run multiple brands in HubSpot, this is the parity piece that was missing. Sandboxes are most useful when they reflect production, and the multi-brand gap meant they often did not. Combined with playbooks in sandboxes above, this is the kind of admin-side quiet win that does not make headlines but materially reduces risk on every change a multi-brand Enterprise team ships.

For more details: Sandbox support for multi-brand supported assets private beta.

Key Takeaways and Next Steps

The agentic surface is widening: The Run Agent workflow action is the practical companion to last week's Agentic Automation Builder beta. The long arc is a canvas-first builder where agents are first-class steps. The short arc, available now, is slotting an agent into the linear builder most teams already use. Either way, the testing discipline matters more than before: agents are non-deterministic, and the combination of agent steps with deterministic workflow steps is where small surprises become big ones in production.

Consolidation is the quiet theme: Centralized data model management, Context Home, and the products work inside Breeze Assistant are all the same shape of update. Take a sprawl of settings or click paths and give it a single, conversational, or visual home. The platform is being designed around fewer places to go and more questions to ask. The admin and AI surfaces are converging.

Sandbox parity is closing: Playbooks and multi-brand assets are not headline updates, but together they meaningfully reduce the risk gap between sandbox and production for Enterprise teams. If you manage a complex portal and have been working around either, this is the time to revisit your sandbox strategy.