I'm a marketing consultant at Adapt. Most of the blog posts you read on Adapt.com pass through me on their way to being published.
The hardest part of editorial work is not writing. It's getting fluent in a company's product, voice, and positioning fast enough to be useful. That fluency normally costs weeks of meetings, scattered docs, and back-and-forth Slack threads that nobody wants to reread.
With Adapt, I am able to produce content much faster. The same Slack channel that holds the team's strategy conversations is where I prompt Adapt to do my keyword and customer research, my outlining and drafting, my publishing, and my post-publish maintenance. The whole content lifecycle lives in one shared thread, where my client can give me feedback.
This post is a walk through of that workflow, with real prompts to Adapt, and the published content we produced.
Step 1: Validate the idea before committing to the post
Blog ideas usually start in collaboration with Ashley, Adapt’s CMO. Sometimes they start with a draft a team member wrote, such as Jim, Het, or Jack.
Before any of us writes a sentence, we use Adapt to validate the opportunity. Adapt is connected to Ahrefs, Grain call recordings, and the customer data warehouse. This allows me to ask these three questions and get answers that are grounded in data:
- Is this topic something our reader actually wants?
- Do search trends support it?
- Are customers talking about this problem on calls?
For example, when I began outlining a post about hosting OpenClaw, my first prompt was the following:
Use ahrefs and tell me what are the related keywords and questions people have about hosting their clankers.
Adapt came back with the keyword table, the search volumes, the difficulty scores, and the long-tail questions people are actually typing into Google. "Install OpenClaw" had 1,000 monthly searches at a difficulty of 2. "How much does it cost to run OpenClaw" was a question with real volume. That data shaped the entire structure of the eventual post.
The customer call version of the same loop happens just as often. Ashley regularly asks Adapt things like:
Review last 15 days of grain calls. How often does the ability to share chats or interact with chats with your team come up as something positive and/or differentiating? Provide direct quotes.
Using real customer call data informs what features we should highlight in future content - and can even help inform product decisions. I’ve also prompted Adapt to generate content ideas. The following prompt queried not just internal data sources, but also Adapt’s published content:
Based on the calls from customers and prospects in the last three weeks, what topics would be helpful to publish on the Adapt site that we haven't addressed yet?
Adapt CEO Jim Benton recently wrote a post where every section is anchored to a specific quote from a specific executive. Adapt helped him to enhance the content with customer evidence.
Step 2: Build content through an outline
I never ask Adapt to generate a finished blog post in one shot. Instead, I start with an outline.
Outlines force you to shape the argument before the wording. This way I make sure the post serves the reader. An outline also gives Ashley a place to give feedback on framing before I have spent time polishing prose.
For example, on our AI knowledge base post, I surfaced a trend I noticed on X and prompted Adapt to do keyword research and make an outline.
Adapt came back with a working title, target keywords mapped to each section, a three-pillar framework (collaboration, automation, app), and a list of FAQ candidates targeting the keyword cluster. But something was missing. I wanted a paragraph that could be surfaced as a Google search snippet, so I added one specific instruction:
Include a part where you clearly define AI knowledge base, in a Wikipedia style.
Before Adapt drafted the full post, the research and structure were already worked out in Slack, where the team could help see and shape them.
The clearest end-to-end example is the our explanation of agents-as-a-service. The outline prompt was:
Open with Jensen giving his talk. Note that the audience groaned wondering if this is really a shift. Define agents-as-a-service. Use the phrase "agent as a service architecture" and talk about how models, data, UI surfaces come together. End with a 5 question FAQ.
Adapt produced a full sectioned outline. We argued about build-vs-buy framing in-thread, and rewrote the definition twice. Again, this took place in a shared Slack channel, where the people who needed to see it could see it. Visibility by design.
Step 3: Strengthen the argument with 3rd-party evidence
Once the outline holds up, we layer in third-party support. This includes studies, market data, recent events, or named experts.
For the Shadow AI 2.0 post Adapt read through a BlackFog survey, the Microsoft Work Trend Index, Samsung's ChatGPT data leak, and OpenAI usage data. It then used these references to enhance and reshape the blog post.
I do not have to bounce between Google, the source's website, my notes, and the doc. Adapt pulls and synthesizes these sources for me in the same conversation. The research happens where the writing happens.
Step 4: Adapt creates the working draft in Google Docs
Once the outline and evidence are settled, the next step is a draft. The Slack prompt I use to create it looks like this:
OK, make a Google Doc with "generated, not ready for review" at the top. Write the whole article in there. Add me as an editor, and anyone who clicks the link can leave comments.
Adapt creates the Doc, sets the permissions, and posts the link back to the thread. From there, I edit line by line. Using Adapt helps me to create a strong starting point, but humans still own the final voice.
Step 5: Adapt moves the post into Payload for publishing
After the draft is ready, I tell Adapt to push it into Payload, our CMS:
Let's add it to Payload as a draft. Make me the writer.
Adapt populates the title, slug, author, category, headings, blockquotes, links, and FAQ structure. It reports each field back so I can sanity check it.
It’s a delightful, friction free experience. The agents-as-a-service post went from Google Doc to Payload draft, to "ready for a final review" inside one Slack thread, in under an hour.
Step 6: Editing and QA still happen, with full context
Before we publish, we ask Adapt to scan for typos and obvious issues. Then a human teammate does a final copy edit.
The advantage of running the workflow in Slack is that the editor has the full backstory. When I tagged my teammate on the agents-as-a-service post, he could see the Ahrefs research, the outline iterations, the Jensen video, the build-vs-buy argument, and the rewrite history.
The copy editor doesn’t have to propose fixes in a vacuum or guess at what the post is trying to accomplish.
Step 7: Maintain and improve old content too
Publishing new content is not the finish line. Once a post is live, Adapt helps the rest of the content library benefit from it.
My standing prompt for internal linking is:
Based on our goals in Ahrefs and the content in Payload, what terms should we internally link to have the most impact?
Adapt cross-references the content in the CMS with our keyword targets and proposes keyword rich anchor text in specific posts. I review the changes in Payload and ship them.
We have also wired up scheduled tasks so that Adapt runs weekly ranking updates on key posts and reports them back into the same Slack channel where the post was written. Our entire SEO maintenance loop is automated.
The bigger takeaway
Adapt goes beyond being an AI writing assistant. It’s a shared intelligence platform that helps me produce high quality content with input from the team every step of the way. Adapt helps us decide what to write, improve what we write, move it into production, and strengthen the rest of our content system afterward.
If you create content inside a company, try Adapt. Connect your data sources and CMS, then see how much faster and more aligned your content workflow can become.
