HubSpot Work uses CRM data in three ways. Its tables link rows to real CRM records, such as a client company and its key contacts. Its AI assistant pulls CRM data when it builds a project. Its agents read that data before they act, such as routing a request by deal properties. HubSpot Work is in public beta as of October 2026 and needs a Core, Sales, or Service seat.
If your client onboarding, deal desk requests, or campaign launches still run in spreadsheets, HubSpot Work is HubSpot's answer for bringing that work next to your customer records. It only pays off when those records are accurate. Before you move that work over, learn below how it connects to the CRM and how its agents use properties. Compare it to the projects object and tools like Asana, then clean up the data your agents will read.
HubSpot Work uses CRM data in three places. Tables link rows to CRM records, the AI assistant pulls CRM data when it builds a project, and agents read CRM context before they act. That's the core difference from a blank project tool.
Tables are the main building block of every HubSpot Work project. In a table, you can link CRM records and add a CRM reference field as a column, next to status dropdowns, dates, and document links. One row in a client onboarding tracker can point to the actual client company and its key contacts.
You can set up a project from a custom prompt, a template prompt, or from scratch. The assistant drafts a plan, waits for your approval, then builds the tables, documents, forms, and agents. Documents can use a table as their source, so a playbook or status summary can reflect the live project data.
Forms feed one table each. That keeps intake requests in the same place as the CRM-linked rows they belong to.
HubSpot describes the payoff in two examples. When a deal closes, the onboarding tracker updates, and when a request comes in, an agent routes it using real deal properties. Work's automation reads the same records your sales and service teams update every day, so the quality of those records decides the quality of the output.
HubSpot Work connects to contacts, companies, and deals, along with their interaction history.
|
CRM Data |
What HubSpot Documents |
What to Check in Your Portal |
|
Contacts |
Named on the product page and used as CRM lookups in HubSpot's onboarding example |
The contact properties that show in a reference column |
|
Companies |
Named on the product page and used as the client lookup in HubSpot's example |
That company associations to contacts are complete |
|
Deals |
Named on the product page, and agents can route requests using deal properties |
The deal properties an agent can read |
|
Interaction history |
Named on the product page as context Work starts from |
How much history an agent uses in a summary or handoff doc |
|
Tickets |
Listed as CRM data that already lives in HubSpot, with no linking steps documented yet |
If a ticket can be a reference column |
|
Custom objects |
Not documented for HubSpot Work |
If your custom objects appear as reference options |
Your AI settings also control what reaches Work. You can manage AI feature access and configure what data is shared in your AI settings. If a Super Admin has restricted data sharing for AI, agents in Work may see less, so check those settings before you judge the results.
HubSpot Work agents read table data and CRM data on a trigger or schedule, then take actions you define. You may use one agent per goal. Its onboarding status agent example reviews new records right away and active onboardings every weekday at 9 AM. It uses table and CRM data to check owners, dates, services, and blockers, then creates subtasks, flags at-risk rows, and notifies the owner.
The documented actions run inside the Work project.
AI steps and rule-based steps are billed differently. AI workflow actions inside a Work agent use HubSpot Credits, while rule-based actions like updating a table value or sending an email notification don't. An agent that runs every weekday will use credits every weekday, so watch usage after launch.
Step 6 is the easiest one to skip. HubSpot's own example tells the agent not to guess when an owner or required field is missing. Our recommendation is to keep that rule in every agent, because a blank owner field quietly breaks any routing logic. The same logic applies to Customer Agent, and our HubSpot Customer Agent setup guide walks through handoff rules in more depth.
HubSpot Work projects and the projects object are two separate features that share a name. HubSpot states plainly that projects in HubSpot Work aren't related to the projects object or Breeze Assistant projects. A Work project is a container for tables, documents, forms, and agents. The projects object is a CRM record type with pipelines, properties, and associations.
|
HubSpot Work Project |
Projects Object |
|
|
What it is |
A container inside a Work workspace |
A standard CRM object you activate in the data model |
|
What it holds |
Tables, documents, forms, and AI agents |
Project records with properties, pipelines, and tags |
|
CRM connection |
Rows link to CRM records through reference fields |
Records associate with tasks, deals, contacts, and other objects |
|
Views |
Table, board, and calendar views inside a table |
Index views, including a Gantt view |
|
Automation |
Work agents built from a prompt |
HubSpot workflows and pipeline rules |
|
Reporting |
Status summaries generated from table data |
Single-object and custom reports |
|
Setup |
Super Admin enrolls in the public beta |
Super Admin activates the object, available on all plans |
|
Status |
Public beta |
Generally available |
Use the projects object when you need reporting and pipeline stages tied to deals. HubSpot's own service delivery example creates a project record from a closed-won deal, so the deal pipeline stays focused on sales.
Use HubSpot Work when the job is coordinating a cross-team process with playbooks, intake forms, and an agent that chases owners.
Many teams will end up using both. A project record can track delivery for reporting, while a Work project holds the onboarding playbook, questionnaire, and status agent around it. Pick one place for each type of data so your team doesn't update the same status twice.
HubSpot Work starts with your CRM data, while standalone tools like Asana and Monday start blank and need an integration to see customer records.
You don't have to pick a side today. HubSpot added Breeze Assistant connectors for tools like Asana, so your team can pull insights and coordinate across both while you test Work. Teams are already adopting HubSpot AI agents for 24/7 support, and Work brings that same agent model to internal operations.
Clean up duplicates, owners, associations, and the properties your agents will read before you turn on HubSpot Work. Every Work agent acts on CRM data, so gaps in that data turn into wrong routes, missed deadlines, and noisy alerts. HubSpot's own numbers show the gap, with customers using AI on high-quality context generating 5x more MQLs than those working from low-quality context.
This is the same lesson we see across onboarding projects. One of the biggest factors that can delay onboarding is having data that isn't ready for migration or implementation. HubSpot Work raises the stakes, because an agent will act on bad data. If your portal needs a reset before you add agents, our HubSpot onboarding and implementation process includes data import, custom properties, and record associations.
Any user with an assigned Core, Sales, or Service seat can use HubSpot Work once a Super Admin opts the account into the public beta. HubSpot lists Work as available on Starter, Professional, and Enterprise tiers of Marketing Hub, Sales Hub, Service Hub, Data Hub, and Content Hub, plus Professional and Enterprise tiers of Smart CRM and Revenue Hub.
Access flows down from the workspace by default. Workspace admins can change anything, and members can view and contribute. Inside a project, owners control the setup, contributors can create and edit assets including agents, and viewers can only look. You can also make a single project or asset invite-only.
Our recommendation is to keep agent creation with a small group of contributors at first. An agent can change table values and message owners on a schedule, so treat contributor access like workflow edit access.
Campaign Creators helps you prepare the CRM data, guardrails, and handoff rules that HubSpot Work agents depend on. The risk with Work isn't the setup wizard. It's an agent acting every morning on duplicate records, missing owners, and loose properties. Our team audits that data first, then scopes the use case, configures the agent, tests real scenarios, and trains the people who review its output.
We've been a HubSpot Elite Solutions Partner since 2022, and we follow the same discover, plan, prepare, configure, prove, and scale framework for every agent launch. If you're planning your first Work agent, see how our HubSpot AI agent implementation service works.
HubSpot Work links its tables to your contacts, companies, and deals, then lets its assistant and agents act on that CRM data. The feature is still a public beta, and it's separate from the projects object. Its results depend on the records underneath, so duplicates, missing owners, and loose properties should be fixed before any agent goes live.
Want to know if your portal is ready for agents that act on CRM data? Check how ready your portal is with the free HubSpot AI Readiness Scorecard.