HighLevel Updates: New Features for October 20-24, 2025

The week of October 20 to 24, 2025 brought one of the largest additions to the platform this year. Agent Studio, a visual builder for multi-agent AI systems, arrived with a free build tier, snapshot support, and public APIs on day one. Alongside it came refreshed default styles in the page builder, a new industry training playbook, and a short list of platform enhancements.

Agent Studio: A Visual Builder for Multi-Agent AI Systems

Agent Studio is a no-code and low-code environment for designing, managing, and deploying AI-driven multi-agent systems. Instead of writing orchestration code, you drag nodes onto a canvas and connect them into a complete workflow. It does not force a choice between autonomous AI and predictable automation: one system can blend reasoning agents, deterministic logic, and conditional routing in the same flow, responding dynamically to user input, tool output, or data you already store.

For an agency, that is the difference between selling a chatbot and selling an operating layer: build one intake and qualification system, then reuse it across clients. The builder also shipped with a brand-new visual interface that makes building, testing, and debugging far less painful.

AI Agent Nodes and Sequential Nodes

Workflows are assembled from two node types. The AI Agent Node creates an autonomous agent powered by a large language model, with its own prompt, model, and variables so you can control tone and function agent by agent. Each agent can be given one or more tools: web search for real-time data, knowledge base search for internal or connected context, an MCP client for secure Model Context Protocol connections, GenAI tools for text, images, video, or audio, and external API integrations for custom services.

The Sequential Node covers everything that should never improvise. It defines deterministic, rule-based steps for structured actions such as form capture, validation, or API requests. Sequential nodes pass variables between steps and connect directly to AI nodes, so an agent can handle the conversation while rigid logic handles the parts that must repeat identically every time.

Conditional Routing, Knowledge Bases, and Runtime Variables

Instead of a router node, Agent Studio puts routing on the edges themselves. You define expressions that evaluate variables or tool outputs to decide the next step, and both expression-based and AI intent-based routing are supported. That gives context-aware branching without manual code, so a pricing question and a support question travel different paths automatically.

Retrieval-augmented generation runs through knowledge bases, and several can connect to a single system. Sources include files, CSVs and tables, websites, FAQs, and formatted text, each with its own conditions so the agent pulls the right material at the right moment. Runtime variables tie it together: values extracted from user input, model responses, or tool outputs can be reused in prompts, routing conditions, and handoffs between agents. Global configs store constants, environment variables, and API keys, while runtime inputs such as a location identifier, a user identifier, or session data pass in as strings, numbers, or booleans. A global prompt sets tone, behavior, and safety rules for every agent so brand voice stays consistent.

Testing, Version Control, and What It Costs

Version history lives inside the builder. Publish to production with one click or revert instantly to a previous stable version, which makes iteration less risky when a client depends on a live system. Before publishing, the whole workflow can be tested in the studio, validating prompts, tool calls, and variable logic in one environment. Deployment happens through Ask AI for real-time conversations, with more integrations and publishing destinations planned.

Pricing rewards experimentation. The free tier covers unlimited agents, workflow execution, knowledge base use, and non-AI features. You pay only for large language model token consumption at public API rates, so prototyping and client demos carry effectively no platform cost. The wider AI toolset shows how this sits alongside existing automation, and our AI feature guide covers the rest of the stack.

Snapshots and Public APIs Extend Agent Studio

Two follow-on releases made Agent Studio far more useful commercially. It is now an asset type in account snapshots, so agents can be captured and loaded, with their associated assets, into any sub-account inside or outside your agency, or distributed through marketplace apps. That turns a well-built agent into a productized deliverable that deploys in minutes instead of being rebuilt for every client.

Public APIs went live at the same time. List Agents retrieves all active agents for a location, Get Agent fetches details for a specific agent by ID, and Execute Agent runs an agent and returns a complete JSON response. OAuth authentication is supported, and private integration tokens are live too, so agents plug into connected environments for end-to-end automation.

Page Builder Elements Get New Default Styling

Default styling for all page builder elements has been refreshed. Headline and content fonts have been swapped for new typefaces, new default images and updated placeholder content ship with every element, and two new colors, Cobalt and Smoke, have been added. The change appears under Page Builder, Add Elements, All Elements, and applies to each newly added element.

The benefit is speed: pages start cleaner and more consistent, so fewer manual adjustments are needed before a page looks professional. Our feature overview and this funnel builder comparison add context.

New Barbershop Training and What Is Coming Next

The Barbershop Playbook, a nine-module guide, joined the training library. It covers account foundations, service calendars, automations that keep clients showing up, a first offer form, collecting five-star reviews automatically, reactivation workflows, and promotions tracked through campaigns and pipelines. Shops that rely on walk-ins and word of mouth have unpredictable revenue, so the material shows how to systemize acquisition and retention. It sits in the University community learning tab, and more industry playbooks follow the same pattern.

Two features were confirmed as coming soon: a custom fields redesign adding date and time, URL, user, calculation, and synced field types, and a rentals calendar. Further out, the roadmap includes AI-assisted refinement of individual workflow actions, AI page building and editing in funnels and websites, mobile app version 4.0, an email domain warm-up tool, and multi-brand support at the agency level. Our beta watch tracks these as they ship.

Smaller Improvements Worth Noting

• SEO issue fixing for external websites is now live for users.
• An SEO pixel installation script has been added.
• Funnels media file migration has been optimized.
• Chat with AI Support is now available on LaunchPad.
• A cleaner node layout, context menus, and improved connection lines arrived with the Agent Studio interface revamp.

Frequently Asked Questions

Does Agent Studio cost extra to use?

The free tier covers building unlimited agents, executing workflows, and using knowledge base and non-AI features. The only charge is large language model token consumption at public API rates, so cost scales with AI usage.

Can one agent be reused across multiple client accounts?

Yes. Agent Studio is supported in account snapshots, so agents can be packaged with their associated assets and loaded into any sub-account inside or outside your agency, or made available through marketplace apps.

What is the difference between an AI Agent Node and a Sequential Node?

An AI Agent Node runs on a large language model and handles reasoning, generation, and contextual responses using attached tools. A Sequential Node executes deterministic, rule-based steps such as form capture, validation, or API requests, for work that must behave identically every time. Both can be combined in one workflow.

Try These Features Yourself

Agent Studio rewards early hands-on time while building and running workflows costs nothing beyond token usage. Spin up an agent, connect a knowledge base, and test a routing condition before publishing. Start a free 30-day GoHighLevel trial to build your first multi-agent system, and work through the onboarding bootcamp if you are setting up a new agency account.