HubSpot has spent years investing in making our platform an incredible place for advertising, gross sales, and repair groups to work. With AI, we have now been constructing AI to do our jobs. for Determine leads, resolve tickets, save offers, and drive outcomes throughout your online business by means of your brokers. This is the reason we name HubSpot an agent buyer platform.
Nevertheless, the agent doesn’t click on on the dashboard or navigate the interface. Name APIs, learn structured output, and carry out actions. Software program constructed for people should evolve to be actually accessible to brokers as effectively.
Nevertheless, entry alone isn’t sufficient. Brokers additionally want substances. Brokers reasoning from uncooked data don’t have any method of understanding what’s regular for a selected enterprise or what has labored for tons of of 1000’s of comparable companies. As I wrote lately, the true AI race isn’t about fashions or information. It is a matter of context.
That perception shapes every little thing we construct. That is why we first shipped MCP Server, and why we have continued to develop what brokers can learn, write, and manipulate ever since. That was just the start.
We’re aiming for a fair larger imaginative and prescient. agent can run above hubspot. And the agent run hubspot.
working above HubSpot signifies that all brokers, whether or not ours or others, can connect with HubSpot’s information, context, and performance as a constructing block. working HubSpot means brokers can work together with the platform end-to-end by means of APIs, MCP servers, CLI, and no matter entry methodology comes subsequent.
what we’re opening up
For brokers to run on HubSpot and for HubSpot to run, they want what’s known as a progress context. It is the particular, dynamic understanding that AI must drive actual outcomes throughout go-to-market, taking into consideration every little thing about an organization’s enterprise, groups, processes, and clients, and powered by patterns throughout HubSpot’s community of greater than 280,000 clients.
It derives from two issues: information and intelligence, and we open each to our ecosystem of shoppers, companions, and builders.
of information layer Contacts, corporations, offers, conversations, tickets, and actions are the muse, open and accessible, and energy 1000’s of integrations at the moment. As at all times, bringing your information into HubSpot is free. Your information belongs to you. Even when they select to go away, it is with them.
of intelligence layer That is what we’re constructing now. It covers each the insights that inform choices (scores, scores, and benchmarks that may be known as immediately) and the actions that drive outcomes (determine leads, resolve tickets, save offers). That is the work Breeze brokers are already doing inside HubSpot, and it is instantly out there wherever your groups and brokers function.
Take buying and selling intelligence, for instance. Gross sales managers pull their group’s open pipeline (quantity, stage, completion date, final exercise) into LLM and ask what’s in danger. This mannequin can calculate averages from pre-prepared information, however it would not inform us whether or not 30 days in-stage is quick or gradual for this trade. The corporate is unaware that these buying and selling champions have gone silent after the reorganization. The corporate isn’t conscious of comparable transactions in comparable corporations stalled final quarter because of this very objection.
With the intelligence layer, a single API name returns a pre-computed threat rating constructed on patterns throughout HubSpot’s tons of of 1000’s of shoppers. You can see that the gross sales cycle on this trade is 90 days as an alternative of 30 days. We additionally see that the champions went silent after the reorganization. We’ll see that different offers like this have stalled over the identical objections earlier than. You may then act on that intelligence by recommending subsequent steps, flagging offers for evaluation, and initiating follow-ups.
The information layer offers the uncooked materials to the brokers. The intelligence layer offers a head begin that can’t be replicated in standalone fashions or in platforms with out networks of this scale.
Ideas on our platform
There are various modifications occurring within the trade proper now. Some platforms reply by closing down, constructing walled gardens, proscribing entry, and making it more durable for patrons to learn from AI. We consider that the other is now required. Openness and belief are extra essential than ever as brokers are empowered to entry information, act on behalf of shoppers, and execute enterprise processes.
Buyer worth above all else. We consider our clients ought to have the liberty to decide on the perfect brokers, integrations, and companions to assist them develop. We’ll at all times spend money on HubSpot’s world-class first-party brokers. However the perfect brokers for specialised industries and workflows typically come from ecosystems. we welcome it.
Open by design. We’re working in the direction of a easy commonplace: something you are able to do inside HubSpot may be carried out by means of the API. Our intelligence needs to be delivered to you wherever you’re employed, inside or outdoors of HubSpot, both immediately or by means of the apps and brokers constructed on prime of us. That is why we’re dedicated to giving builders entry to the identical foundations we construct on.
Trusted by default. We deal with belief and governance as core infrastructure. When clients connect with accomplice instruments, launch brokers, or construct one thing customized, they should know precisely what they’ve entry to and what they’re doing. An agent appearing in your behalf is simply helpful should you belief it.
These aren’t simply ideas. These are intentional selections of what sort of platform we wish to be.
What’s presently out there and what’s coming subsequent?
Now: An open and agent-enabled platform. HubSpot is presently open to brokers. Our API and MCP servers are up and working. Claude, ChatGPT, Gemini, and Copilot connectors provide actual worth to our clients. Over 2,000 apps are working throughout our ecosystem, and new brokers are constructed on the platform each week.
Subsequent comes full API parity. We proceed to develop our public API floor so that each characteristic of the platform – each workflow, each motion, each context – is accessible to apps and brokers constructed on prime of it. No performance ought to exist solely behind the UI.
Alternative for the long run
The shift to brokers is already occurring with each GTM group making an attempt to determine the long run path of their work, and with each builder deciding which platforms are value investing in. We consider the reply comes right down to context. An ideal agent is one who understands enterprise the identical method an incredible marketer, salesperson, or CSM would: what’s regular, what works, what’s altering, and what has labored for these corporations.
That is what HubSpot has spent 20 years creating for greater than 280,000 corporations. And we open up that intelligence to each agent, each accomplice, each buyer to form what occurs subsequent. We do not assemble all of the solutions. However we construct what we want for each reply.
