Build The AI Agent. HighLevel's Ask AI Does The Rest
Share this post!
If you are on HighLevel, there is a reasonable chance you are using about a third of what it can actually do. That is not a dig. The platform is broad, time is short, and most people get the core set up and keep moving. Which works, until you start to see what is sitting there unused.
Agent Studio has been in HighLevel for a while now, letting you build AI agents that live inside your CRM and carry out work on your behalf. Ask AI has just had a significant upgrade that changes what is possible with those agents, because you can now map them directly into Ask AI and call them from a single conversational interface. Build the agent once, describe its capabilities, and from that point you just ask. Ask AI figures out which agent to trigger, collects what it needs, and gets it done.
Even people who work with HighLevel daily are still working out the full extent of what these two things can do together. That is not a weakness of the platform. It is just a genuinely new way of working, and the possibilities are still being discovered. This post covers what each feature does, what you can realistically do with them, and how to get started.
Agent Studio: your AI workforce inside HighLevel
Agent Studio is where you build AI agents that live inside your HighLevel account and do work on your behalf. Not respond to customers the way a basic chatbot does, but actually carry out tasks: gathering information, making decisions, generating content, sending things, scheduling things, looping back when something needs a second look.
The meaningful difference between an agent and a standard automation workflow is that agents can handle variation. A workflow sends the same email every time regardless of context. An agent reads the situation, asks follow-up questions if it needs to, takes a different path based on what it gets back, and only escalates to a human when it genuinely cannot handle something itself. That distinction matters a lot once you start thinking about what you would actually want to hand off.
A few examples of what that looks like in practice:
Free website audit as a lead magnet
You offer a free audit to generate enquiries. Normally each one takes an hour of your time. An agent can carry out the audit autonomously and deliver a written report to the prospect without you touching it. You could offer this at any volume without the time cost scaling with it.
Social media content with an approval loop
An agent generates a caption, image and hashtags for a given topic, then checks whether you are happy with it. If not, it generates a fresh set. Once you approve it, it schedules the post. You stay in control of what goes out without doing the creation work yourself.
Inbound leads outside of office hours
Someone gets in touch on a Friday evening. Rather than sitting in an inbox until Monday, an agent handles the initial conversation, qualifies them against your criteria, updates your pipeline, and flags anything urgent. By the time you look at it, the lead is warm and documented.
Client onboarding
An agent walks a new client through your onboarding process, collects the information you need, handles standard questions, and only passes to you when something genuinely requires your input. What used to be several rounds of back and forth emails largely takes care of itself.
The thread running through all of these is not that AI is doing something impressive. It is that specific tasks which currently sit on your plate could sit on an agent’s instead, and you would not need to be involved unless something went wrong or a human judgement was needed.
Three ways to build an agent
There are three routes into Agent Studio depending on how you like to work and what you are building.
The first is building from scratch manually. You start with a blank canvas and configure the logic, tools and flow yourself step by step. This gives you full control and suits more complex or specific builds.
The second is using Build With AI, a natural language assistant inside Agent Studio. You describe what you want in plain English, it plans the agent back to you for approval, and then builds it. You end up with a properly structured agent without having to configure each step manually. The example below uses this method.
The third is starting from a template. Agent Studio has a library of pre-built agent structures covering common use cases. You install one with a click, customise it for your setup, and you are ready to go. The fastest route if you want something standard up and running quickly.
What building an agent actually looks like
Here is a real example using the Build With AI method. The brief was a social media content agent that generates a full post package, checks for approval, regenerates if needed, and schedules once approved. The prompt used was this:
“Create a social media agent that generates a caption, image and set of hashtags according to the topic given. The agent should check if the user is happy with the content. If the user is, the agent should proceed to scheduling the post according to the given time and date. If the user is not happy with the post, the agent should generate a fresh post, including copy, image and hashtags.”
Build With AI planned the agent back before building anything: it would ask for a topic, platform, tone and intended publish date, generate a caption, image and hashtags, ask for approval, regenerate the full package if rejected, and schedule once approved. It also suggested adding platform selection so caption length and hashtag style could be matched to the channel.
One click to approve the plan. It built the agent.
A complete content creation and scheduling workflow, human approval loop included, from a two sentence brief. That agent now sits in the account ready to run whenever it is needed, and can be called directly from Ask AI without ever opening Agent Studio again.
Ask AI: where you talk to everything you have built
Ask AI is HighLevel’s built-in assistant, sitting in the top navigation bar with reach across your whole account. On its own it is already useful: it can create branded content using your Brand Voice settings, set up communities, handle CRM actions like creating records or rescheduling appointments, create Voice AI agents through a guided chat, and help with campaign ideas and copy. All from a single prompt window without navigating through menus.
The recent upgrade is the ability to map your Agent Studio agents into Ask AI directly. Once an agent is mapped, it becomes part of Ask AI’s toolkit. You do not go and find it, open it, or trigger it manually. You just describe what you need, and Ask AI works out which agent to run, collects any information it needs at runtime, and brings the result back in the same conversation.
To use the earlier example: you have built the social media content agent. You open Ask AI and type something like “create an Instagram post about our spring offer.” Ask AI recognises this matches your content agent, triggers it, asks for the details it needs, runs the workflow, and returns the post for your approval. The agent is doing the work. You just asked.
Why this matters:
Before this upgrade, your agents lived in Agent Studio and had to be triggered or managed from there. Now they are accessible from a single conversational interface alongside everything else Ask AI can do natively. One place, plain English, and Ask AI routes intelligently to whichever agent or built-in capability is the right fit for what you have asked.
How to map an agent into Ask AI
Once an agent is built in Agent Studio and promoted to Production, connecting it to Ask AI is straightforward:
- Go to Agency Settings and find Ask AI Configuration.
- Click Map New Agent.
- Select the sub-account where the agent lives and choose the agent.
- Add a description of what the agent does and list its key capabilities. These are what Ask AI uses to decide when to trigger it, so the clearer you are, the smarter the routing.
- Map any input variables the agent needs: either pull from a custom field automatically, or set it to ask the user at runtime.
- Click Save and Integrate. The agent is now live inside Ask AI.
Mapping happens once at agency level and applies across your sub-accounts. Ask AI manages your custom agents alongside its own built-in capabilities, and you can add, update or remove mappings from the Ask AI Configuration table at any time.
One thing worth noting: the capability descriptions you write during mapping are doing real work. Ask AI uses them to decide which agent is the right fit for a given request. A vague description produces vague routing. A specific, accurate description of what the agent handles means Ask AI will know exactly when to call on it.
Finding Ask AI and Agent Studio in your account
Both features are available within HighLevel. If you cannot find them, they may need to be switched on in Labs.
Agency owners and admins:
Both Ask AI and Agent Studio should be accessible from your dashboard. If not, go to Agency Settings > Labs and enable them. You can also enable them per sub-account via Labs in the sub-account settings.
Sub-account users:
You will need your agency admin to enable these features for your account.
Direct HighLevel subscribers:
Check your Labs settings to enable both features.
Do this before anything else
Set up Brand Voice. Both Ask AI and agents built in Agent Studio pull from it when generating content. Without it, what comes back will be competent but generic. Something that needs rewriting before it sounds like your business. With it configured, you are starting from content that already has your tone, your terminology, and your style.
You will find it under Marketing > Branded Content > Brand Voice. It takes about ten minutes and it is the kind of thing that makes a noticeable difference to everything downstream without you ever having to think about it again. If you are investing time in building agents and setting up Ask AI, do Brand Voice first.
Need support for your business?
Get in touch at onyxops.co.uk
OnyxOps provides business and operations support for small businesses. Hands-on admin and marketing, automation and AI integrations. Whatever is eating your time or holding your business back, that is where we come in.