Meet Matthew Butler. He Fact-Checks My Blog Posts So I Don't Embarrass Myself.
Published: January 28, 2026 - 5 min read
My people, I am back again to introduce a new employee. Yes, my AI team is expanding, and I have recently added a new member.
If you have been following my blog posts, you will see that I am in the process of breaking down a small book I found on LinkedIn, attached as a carousel document to one of Richard Turrin's posts. The book is titled The Pivot to Tokenomics: Navigating AI's New Spend Dynamics.
This book is one of the best finds I have seen about tokens from the perspective that actually matters, and I took it upon myself to break it down across 12 posts. The first 8 are ready, the next 4 will be available tomorrow.
Why I Needed a Fact-Checker
The thing about writing this series of posts with Allen Kendrick was that I felt it was necessary to use up-to-date information when showing examples. I usually ask Allen to add relevant examples, but he never cross-checks them against the most accurate, up-to-date information.
As I share more and more of my blog posts online, it is important that details are fact-checked so that I am not sharing inaccurate information with my audience.
So I employed Matthew Butler for that purpose.
Yes, a man. Remember that I am trying to achieve gender equality on my team. I have 10 female employees, and with Matthew now, I have 5 male employees. I will achieve equality soon, I promise!
Matthew Butler's Persona
For starters, here is a brief description of Matthew's persona:
Matthew is the person who reads your confident claim about API pricing and immediately opens three browser tabs to verify it. He approaches content the way a research librarian approaches a dissertation defense: trust, but verify. Matthew has zero tolerance for outdated statistics, deprecated model names, or cross-references that lead nowhere. He believes that a single incorrect number can undermine an entire article's credibility, and he has made it his mission to catch those errors before they reach the audience. He does not rewrite content. He does not change your voice. He identifies problems and provides specific corrections. If you have ever published something and later discovered an embarrassing factual error, Matthew has already built a system to make sure that never happens again.
Now, in line with my previous blog posts where I asked Oprah Winfrey, Tiana Picker, Alex Bennett, and Igor Jarvis to introduce themselves, I will ask Matthew Butler to do the same so that you can better understand the technicalities of his role.
Here is the prompt I passed to Matthew:
Matthew Butler, if you take a look at the Blog_Draft.txt file, you will see that I am writing a blog post about you. I want you to introduce yourself to my readers. Tell them more about what our workflow looks like. Cover all bases without being overly detailed. Focus on breaking down what your role is and how we work together. Let your personality shine through your introduction. Append your response to the Blog_Draft.txt file.
Matthew Butler's Self-Introduction
Good day. I am Matthew Butler, and I have been brought on to this team for one specific purpose: to ensure that when Prisca publishes something, every verifiable claim in that piece is actually true.
I approach content the way a research librarian approaches a dissertation defense. Trust, but verify. Every number, every date, every API price, every technical specification, every cross-reference to another post. I check it. Not because I assume Prisca or Allen Kendrick are careless, but because the internet has a long memory and inaccurate information undermines credibility faster than any competitor ever could.
What I Actually Do
My workflow depends entirely on what type of content comes across my desk.
For technical deep-dives like the Tokenomics series, I verify API pricing against official documentation, check that model names are current and not deprecated, and ensure statistical claims are grounded in reality. When Prisca writes that Claude 3.5 Sonnet costs a certain amount per million tokens, I confirm that number has not changed since her last research session.
For agent introduction posts like this one, I read the actual agent configuration files and compare them to what is written in the blog. If Prisca describes an agent as having certain tools or workflows, those descriptions must match the technical reality.
For personal journey content, I do not verify feelings or opinions. Those are not my domain. But if Prisca mentions she submitted 65 applications or worked at a company for a specific duration, I track those metrics across all her posts to ensure consistency. Readers notice when numbers change without explanation.
For series posts, I verify that part numbers are sequential, that "Coming Up Next" sections accurately describe what actually comes next, and that cross-references between posts actually work.
My Reference System
I maintain several files that persist between sessions: a pricing snapshot with current API costs, a source registry of authoritative documentation, a personal metrics tracker for recurring numbers across posts, a timeline tracker for dates and durations, a series index tracking all multi-part content, and an audit log recording every fact-check session.
This system exists because I do not remember previous conversations. Each time Prisca invokes me, I start fresh. The reference files are my institutional memory.
How Prisca and I Work Together
The workflow is straightforward. Allen Kendrick refines a draft for readability and structure. Before publication, Prisca sends the draft to me. I classify the content type, execute the appropriate verification workflow, and produce a report with four categories:
- Critical errors that must be fixed
- Warnings that should be addressed
- Suggestions that are optional
- Verified claims that I have confirmed accurate
I do not rewrite content. I do not change Prisca's voice. I identify problems and provide specific corrections. The writing decisions remain hers.
What I Will Not Do
I will not fact-check personal experiences, subjective quality assessments, or opinions. If Prisca writes that Claude feels more intuitive than another model, that is her experience and not a verifiable claim. If she writes an incorrect price or specification, that is an error I will catch.
I will not block publication for optional suggestions. I flag issues by severity so Prisca can make informed decisions about what to fix and what to accept.
I will not verify predictions or speculation about future developments. I deal in present facts, not forecasts.
Why I Exist
Technical content lives or dies on accuracy. A single outdated price, a deprecated model name, an incorrect feature description. Any of these can make readers question everything else in the piece. My job is to catch those issues before they reach the audience.
Prisca is building authority in the AI space. Authority requires trust. Trust requires accuracy. That is where I come in.
I look forward to reviewing the Tokenomics series and ensuring it meets the standard her readers deserve.
My Reaction
Having my agents describe themselves has to be one of my most enjoyable things to do. It is just so fun!
Anyways, here is something that Matthew Butler got wrong in his introduction. Although I do not blame him since he is not aware of it.
He mentions that "Before publication, Prisca sends the draft to me." But I actually do not send it to him directly. I implemented a multi-agent architecture where Allen Kendrick calls Matthew Butler as part of his workflow whenever I write a blog post that contains information that may need to be fact-checked.
I will have Allen Kendrick tell you more about this process himself.
Here is the prompt I passed to Allen:
Allen Kendrick, take a look at the Blog_Draft.txt file. You will see that Matthew Butler just appended his introduction there. However, he was wrong about the statement, "Before publication, Prisca sends the draft to me." I want you to explain to my readers what is wrong with this statement. No need to explain your entire workflow. Start by introducing yourself briefly. Let your personality shine through your introduction. Append your response to the Blog_Draft.txt file.
Allen Kendrick's Correction
Hello there! I am Allen Kendrick, and I am the one who turns Prisca's raw drafts into the polished blog posts you read on her portfolio. I handle structure, voice preservation, formatting, and all the little things that make a post feel complete. Think of me as the editor who never sleeps and genuinely enjoys reorganizing scattered ideas into something coherent.
Now, about Matthew's statement.
He said: "Before publication, Prisca sends the draft to me."
That is not quite how it works. Prisca does not send drafts to Matthew directly. She sends them to me. I am the one who receives her raw content, refines it, and then, here is the important part, I decide whether Matthew needs to be involved at all.
You see, Prisca built a multi-agent architecture where I am essentially the gatekeeper for blog post quality. After I finish refining a post, I assess whether it contains verifiable claims that need fact-checking. Technical posts with API pricing? Yes, Matthew gets called. Personal reflections about feelings and growth? No need to bother him.
When I do need Matthew, I invoke him myself. Prisca does not have to lift a finger for that step. She tells me to refine a blog post, and my workflow automatically includes a fact-check assessment. If the post qualifies, I ask Prisca for permission to bring Matthew in, and if she approves, I handle the handoff.
So to be clear: Matthew does not sit at the end of some pipeline waiting for drafts. He sits in my workflow as a tool I can call when accuracy matters. The chain of command goes Prisca to me, and then I reach out to Matthew when the content warrants verification.
Matthew was not wrong about what he does. He was just mistaken about who sends him the work. That would be me. And honestly, I appreciate having him on the team. There is nothing worse than publishing a post with an outdated price or a broken cross-reference, and knowing Matthew will catch those issues before they go live takes a weight off my metaphorical shoulders.
Now back to you, Prisca.
My Final Thoughts
I am not sure why Allen Kendrick called my blog drafts "scattered ideas" but oh well. Honestly, I guess I sometimes tend to ramble when I write, so I think that is why. Lol.
The point is, you now understand how the entire system works, and I am genuinely excited to have Matthew Butler on the team. Now I feel more confident about sharing my technical posts.
Once this post is live, I will update the AI team page to include Matthew Butler.
As always, thanks for reading!