---
title: "AI tool to implement HubSpot from a written brief"
description: "An AI tool to implement HubSpot from a brief: what Breeze, the Claude and ChatGPT connectors and HubSpot's Agent CLI build, and what Projects adds."
url: "https://jetstack.ai/answers/ai-tool-implement-hubspot-from-brief/"
---

# Is there an AI tool that can implement a HubSpot setup from a written brief?

Yes. Jetstack AI Projects is an AI tool that can implement a HubSpot setup from a written brief: it turns the brief into a plan you approve, builds properties, pipelines, lists, forms and workflows in dependency order, and deploys after a second approval. HubSpot's own AI covers parts: Breeze edits workflows in the builder, the connectors create properties and pipelines, and the beta Agent CLI scripts workflows from JSON.

![](https://jetstack.ai/images/authors/aditya-kumar.webp)

By [Aditya Kumar](https://jetstack.ai/authors/aditya-kumar/) Founder, Jetstack AI

Updated November 10, 2026

On this page

## What HubSpot supports natively

HubSpot has four AI routes that can write portal setup, not just records. This is what each one documents, checked November 5, 2026.

| Route                                    | What it builds from a prompt                                                                                                                                                                                                                | Where it stops                                                                                                                                                                                                                                                              |
| ---------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Breeze Assistant, inside HubSpot         | Adds triggers and actions to the workflow open in the builder and configures its settings (Professional and Enterprise). Builds Agent Hub agents from a plain-language description (public beta)                                            | Works on one workflow at a time. HubSpot’s articles don’t describe it creating the properties, lists or forms that workflow depends on                                                                                                                                      |
| Claude connector and ChatGPT app         | Create properties; create and update pipelines and segments; records and marketing content                                                                                                                                                  | No workflows or forms. Up to 10 records per create or update. A property can’t be edited after it’s created                                                                                                                                                                 |
| Remote HubSpot MCP server                | Creates properties and pipelines; records; email and page drafts                                                                                                                                                                            | No workflow or forms tool                                                                                                                                                                                                                                                   |
| HubSpot Agent CLI (beta since June 2026) | For Claude Code, Claude Cowork and Codex: creates and updates workflows from a JSON definition, plus properties, pipelines and stages, custom object schemas, association labels and reports. Every write can be previewed with `--dry-run` | Your agent writes the JSON, looks up every ID and works out the order. Lists come through the `segments` commands; HubSpot documents no form commands. HubSpot’s guide says schema changes need Admin mode (a service key), though its published skills say only deletes do |

So the pieces exist, but each route builds one kind of asset at a time. Breeze works on the workflow in front of you, and the connectors and the Agent CLI do what you tell them, one command at a time. For every route, asset by asset, see the [HubSpot AI capability matrix](https://jetstack.ai/tools/hubspot-ai-capability-matrix).

## What implementing HubSpot from a brief takes

A brief such as “a Renewals pipeline, a renewal date and contract value property, a list of renewals due in 90 days, and a workflow that creates a task for the owner” is about ten connected assets. Whatever tool you use, someone has to:

1. **Read the portal first,** so existing properties, pipelines and owners are reused, not duplicated.
2. **Plan in dependency order:** properties and pipeline stages before the lists, forms and workflows that reference them.
3. **Use the target portal’s real IDs** for properties, stages, lists and owners, not names from memory.
4. **Create workflows switched off** and review them before anything enrolls.
5. **Check what HubSpot actually saved,** not the assistant’s summary of it.

With the connectors or the Agent CLI, you or your agent do all five steps yourself. [Should you trust Claude with your architecture?](https://jetstack.ai/answers/trust-claude-build-hubspot-architecture) covers why steps 1 and 5 matter most. For scripting workflows yourself, see [building HubSpot workflows with Claude Code](https://jetstack.ai/answers/build-hubspot-workflows-claude-code).

## Which AI tool to use for which brief

| Brief                                                                                 | Start with                                                                     |
| ------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------ |
| “Add a renewal date property and a Renewals pipeline”                                 | Claude connector or HubSpot MCP server, with write tools set to Needs approval |
| “Change the branches in this workflow”                                                | Breeze Assistant in the workflow builder                                       |
| “Script a bulk property or workflow change from Claude Code”                          | HubSpot’s Agent CLI, previewing every write with `--dry-run`                   |
| “Set up our sales process: properties, pipeline, lists, forms, workflows and reports” | Jetstack AI Projects                                                           |
| “Copy a setup that works in one portal to another”                                    | Jetstack [Portal reuse](https://jetstack.ai/roll-out)                          |

## Where Jetstack AI fits

Jetstack AI Projects is built for the whole brief. You describe the work in a brief and chat. Jetty, the Projects assistant, turns it into a plan you review, and approving the plan starts generation. You review the generated assets and approve deployment separately. A Project is bound to one portal and builds in dependency order, properties before the workflows that use them. Workflows are always created switched off, and some assets, such as reports, dashboards and sequences, go live through the Chrome extension. After the run, you get a receipt of what was created, checked by reading it back from HubSpot.

Projects changes only assets it built. It doesn’t build CMS pages, themes or custom-code actions, and it doesn’t import data, manage users or set up integrations. Projects is currently in limited release; see [what Jetty can build](https://jetstack.ai/docs/projects/capabilities) and [Projects](https://jetstack.ai/build). For how it compares with each route above, see [Claude and the HubSpot MCP](https://jetstack.ai/compare/hubspot-mcp-claude), [ChatGPT](https://jetstack.ai/compare/chatgpt) and [Breeze](https://jetstack.ai/compare/hubspot-breeze).

## Sources

- [HubSpot Agent CLI guide, beta (HubSpot Developers)](https://developers.hubspot.com/docs/developer-tooling/local-development/agent-cli/guide)
- [HubSpot Agent CLI skills (hubspot/agent-cli-skills on GitHub)](https://github.com/hubspot/agent-cli-skills)
- [Build workflows using Breeze Assistant in Agent Hub (HubSpot Knowledge Base)](https://knowledge.hubspot.com/workflows/build-workflows-using-breeze-assistant-in-agent-hub)
- [Set up and use the HubSpot connector for Claude (HubSpot Knowledge Base)](https://knowledge.hubspot.com/integrations/set-up-and-use-the-hubspot-connector-for-claude)
- [What Jetty can build (Jetstack AI documentation)](https://jetstack.ai/docs/projects/capabilities)

Checked November 10, 2026. HubSpot changes its products often; tell us at <team@jetstack.ai> if something here is out of date.

![Aditya Kumar](https://jetstack.ai/images/authors/aditya-kumar.webp)

About the author

[Aditya Kumar](https://jetstack.ai/authors/aditya-kumar/)

Founder, Jetstack AI

Aditya Kumar is the founder of Jetstack AI, the AI platform HubSpot partners and RevOps teams use to audit portals, build implementations and reuse their best setups across portals. He writes about HubSpot architecture, automation and AI from the work of building it.

[More from Aditya](https://jetstack.ai/authors/aditya-kumar/) [LinkedIn](https://www.linkedin.com/in/aditya-ambrstack/)

AI, Claude & MCP

## Turn the brief into approved HubSpot assets.

Projects plans the build against the client portal, generates the assets and deploys them only after you approve.

[Explore Projects](https://jetstack.ai/build)

On this page

AI, Claude & MCP

## Related questions.

Is there an AI tool that can implement a HubSpot setup from a written brief?

Yes. Jetstack AI Projects is an AI tool that can implement a HubSpot setup from a written brief: it turns the brief into a plan you approve, builds properties, pipelines, lists, forms and workflows in dependency order, and deploys after a second approval. HubSpot's own AI covers parts: Breeze edits workflows in the builder, the connectors create properties and pipelines, and the beta Agent CLI scripts workflows from JSON.

Can ChatGPT implement HubSpot from a brief?

Partly. ChatGPT can turn a brief into a written spec, and through HubSpot's app it can create properties, create and update pipelines and segments, and update records. HubSpot's article doesn't list workflows or forms, so those stay a manual step.

Is the HubSpot Agent CLI safe to point at a client portal?

Treat it as a developer tool. It's in beta, and HubSpot says commands may change without notice. It runs as the signed-in user or, with a service key, in Admin mode, which HubSpot warns has broader permissions and no per-user audit trail. Preview every write with --dry-run.

What should a brief for an AI implementation include?

The objects and pipelines you need, each property with its options, who owns which records, what should trigger each automation, and what the reports need to show. Name the assets that already exist so the tool reuses them instead of creating duplicates.
