Our content is human-led and AI-assisted. We choose the topics, perform the research, pull the data, run the reports, capture the source screenshots, write the notes and examples, choose the page structure, and make the final publishing decisions. AI primarily helps organize, polish, format, and assemble that work before a manual review and quality-control process.
AI is part of our publishing workflow, but it is not the source of our experience, client data, screenshots, examples, opinions, or final decisions. This page explains exactly where human work ends, where AI assistance begins, and what we manually verify before anything is published.
Transparency is more useful than pretending modern content workflows do not use AI tools. Google's own people-first content guidance encourages publishers to make the who, how and why behind content clear when that context helps the reader. This disclosure is our answer to the “how.”
AI normally enters the process after the important editorial decisions have already been made. On larger pages, the source material can include multiple documents, screenshots, reports and many pages of notes before an agent ever sees the project.
The process begins with a real reason to create the page. That might be a client result worth documenting, a recurring question from HighLevel agencies, a workflow we have tested, a business lesson from our agency, or a topic that needs a clearer explanation. We decide what the reader should leave knowing before we decide what the page should rank for.
Human DecisionWe pull the account data, run the reports, capture the screenshots, collect relevant URLs, record the specific metrics we want to explain, identify supporting examples, and organize whatever firsthand information is needed for that page. For strategy content, the source material usually comes from our own agency processes, client work, testing, support conversations, or systems we actively use.
Human ResearchBefore AI assembles a page, we write out the points that need to be made in note, list, outline, or paragraph form. That can be a short briefing for a simple page or dozens of pages of material for a large resource. Some projects have more than 35 pages of notes, screenshots, instructions, examples and data before the page body is assembled.
Human Source CopyThe layout is not left open-ended. We already know the page type, design system, heading hierarchy, content components, image placements, internal-link rules, responsive behavior, and the sections we want. Those structures come from our own experience building HighLevel websites and repeatedly testing what makes long-form content easier to understand, read and maintain.
Human FrameworkOnce the evidence, notes and structure are defined, one of our custom agents can help turn that material into the requested page. Typical assistance includes organizing the supplied notes, removing unnecessary repetition, smoothing transitions, keeping terminology consistent, fitting content into the chosen components, and producing the HTML, CSS or other code needed for the finished body.
AI AssistanceThe generated body is not treated as a finished page. We review the wording against the source material, inspect the code, add or rewrite paragraphs where needed, check the screenshots and links, proofread the page, test desktop and mobile layouts, and run our normal quality-control process. We also write and apply the page's final metadata ourselves. Only after that review is the page considered ready.
Human Final ApprovalOur preference is to make as few important editorial decisions with AI as possible. The more consequential the decision is to accuracy, evidence, strategy, or meaning, the more likely it is handled directly by a person.
The role of AI changes when a page contains client performance data. The numbers and report images have to come from the actual source, not from a generated approximation of what a report could look like.
Our public our published case studies are built from real client work. Business names and private account details may be anonymized, but the underlying report data and screenshots are not recreated with AI.
This disclosure covers the editorial content on this site first. The agents we use for client and partner websites are a separate product workflow, but the underlying idea is similar: AI-assisted production should work from controlled inputs, real business context and human checkpoints.
Topics, research, keywords, business information, custom copy, links, images and other page inputs can be selected or supplied before the page is assembled.
We encourage clients to keep real business photos in HighLevel media storage so agents can select from authentic images first. If no suitable real image exists, a generated image may be used until a better real image can replace it.
A user can provide more of the writing manually and use an agent primarily to organize content into the approved page structure, rather than asking it to generate the substance from scratch.
Faster production is useful only when the page still has something real behind it. These are the parts of the process where firsthand work and human accountability matter most.
The review is both editorial and technical. A useful page can still fail the reader if a number is wrong, a screenshot is misleading, a link is broken, or the mobile layout makes the content difficult to use.
That is the balance we are trying to maintain across this site: use AI tools to reduce repetitive production work without outsourcing the experience, evidence, judgment and accountability that make the content worth publishing in the first place.
