Matt Gray • 2026-06-03
Overview
| Creator | Matt Gray |
| Title | how I'm automating 97% of my business with AI |
| Source | youtu.be/se4rp_VAUMk |
| Transcript Date | 2026-06-03 |
Matt Gray walks through the operating system he has built to automate the large majority of his business using AI, Claude Code, and centralized dashboards. The core philosophy is to shrink headcount, stop replacing departing staff with humans, and default every repeatable process to automation or AI-driven tools instead. Rather than chasing AI for its own sake, Gray frames every automation decision around a single test: does it cut costs or increase profit?
The walk-through covers his "Project Automate" dashboard, marketing and operations machines built in Notion-style environments, an AI orchestrator concept that turns humans into editors rather than doers, and an automated DM and cold outreach system called Alberto that personalizes messages at scale across LinkedIn, X, and email. The overall system is positioned as an ongoing build, not a finished product, with Gray committing to continue sharing updates as the technology evolves.
The session is practical and candid, including an honest admission that even his own businesses face constant operational constraints and gaps. The value is in seeing a real business owner's decision-making logic for where and how to apply AI, rather than a polished vendor pitch.
Key Points
Quotable Moments
Quotable moments are auto-generated from the transcript. Speaker attribution and quote accuracy should be verified against the original source before republishing or sharing.
Matt Gray
"We are shrinking the amount of folks that we have working in the company. We kind of are not hiring much more, and then we are defaulting to automation."
Why it works: This is a concrete strategic position, not a vague aspiration. Gray is stating an active policy decision, which makes the automation talk credible and grounded.
Matt Gray
"We are either reducing costs and or we're increasing profits. That's the only reason to use it."
Why it works: A clean, quotable filter that cuts through AI hype. Most audiences need a practical frame for evaluating technology investments, and this delivers one in a single sentence.
Matt Gray
"Too many people are just vibe coding a bunch... they build all this stuff, but it's not usable. No one on their team knows what's going on."
Why it works: "Vibe coding" is a term that resonates immediately with anyone who has watched AI projects balloon without direction. The critique is pointed but self-aware, coming from someone building in the same space.
Matt Gray
"If humans are on the team, they're doing stuff that only humans can do. We don't have people doing stuff that robots can do."
Why it works: A clear articulation of the human-AI division of labor principle that is easy to repeat and apply. It functions as a hiring and role-design filter.
Concepts and Ideas
Core Framework -- Automation as Operating System
Project Automate Dashboard
A centralized command center that surfaces strategy, weekly actions, Q2 OKRs, Kanban progress, and sprint checklists in one view. The goal is to eliminate the need for status meetings and keep leadership aligned without constant interruption. It functions as the operating layer between strategic intent and daily execution.
Default to Automation
Rather than treating AI as an add-on, Gray has made automation the default response to any new operational need. When a role opens up or a new workflow is needed, the first question is whether a machine can handle it, not whether to hire. This shifts the organizational posture from human-first to automation-first.
The Two-Filter Test
Every automation is evaluated against two outcomes: cost reduction and profit increase. If an automation does neither, it does not get built. This filter prevents the common trap of building impressive but economically inert AI projects that generate engagement but no business return.
Practical Principles -- How the System Works
Orchestrator Model
An AI orchestrator sits at the center of multi-step business workflows. When triggered by a command like "spin up a new offer," it autonomously creates landing pages, offer pages, and email sequences. Humans then serve as editors, reviewing and refining the AI output rather than doing the manual build work themselves.
Humans as Editors
The orchestrator queues tasks for humans to review and approve. This inverts the traditional workflow where humans initiate and AI assists, replacing it with a model where AI initiates and humans quality-control. It concentrates human attention on judgment and approval rather than production.
Shared Knowledge Brain (SDK)
A centralized institutional knowledge base that AI agents across the company draw from. Rather than each AI interaction starting from scratch, the shared brain gives every automated process access to consistent context about the company's voice, offers, and operating logic. This reduces hallucination risk and improves output quality at scale.
Whiteboard-First Build Process
Before building any AI system for a department, Gray sits with the relevant leader, maps every process, document, and spreadsheet in that function on a whiteboard, and identifies the gaps and redundancies. The automation is then built to reflect the real workflow, not an imagined one. This prevents vibe coding and ensures the system is actually usable.
Alberto -- Personalized Outreach at Scale
An outreach automation that identifies people who have already engaged with Gray's content (comments, followers) and sends them personalized DMs in his voice across LinkedIn, X, and cold email platforms like Instantly and Apollo. When prospects express interest, the system routes them to the appropriate portfolio company call booking flow.
Adjacent Frameworks Referenced
FounderOS Velocity
Gray's coaching product that delivers a 30-day brand and content build-out for founders. The onboarding machine behind it is itself an example of the systematized delivery approach he describes -- structured milestones, accountable timelines, and AI-assisted production of brand positioning, content strategy, and email sequences.
Level 10 Meeting Structure
A reference to the EOS (Entrepreneurial Operating System) level 10 meeting format, integrated into Gray's ops machine dashboard. It connects the annual plan to the quarterly plan to weekly execution in a single operating view, giving the team a shared cadence for reviewing progress and surfacing issues.
Content Monetization Stack
The downstream layer of the content machine: newsletter, offer stack, and lead magnets working together to move audience members from social platforms (rented audiences) to email lists (owned audiences) and then to products or services. Gray is replacing traditional lead magnets with AI-powered tools built using Claude Code.
Implementation
Implementation steps are auto-generated from the transcript content and are provided for informational purposes only. They do not constitute professional advice of any kind. Always consult a qualified professional before acting on any information presented here.
Apply the Two-Filter Test to Every Current Tool
Before adding any new AI tool or automation, audit what you already have running. For each one, ask whether it has demonstrably reduced a cost or increased a revenue stream since you deployed it. If you cannot answer that question with a number, the tool is occupying time and budget without justification. Cut or freeze anything that fails the test, and use the freed capacity to fund things that pass it.
Run a Whiteboard Session for One Department
Pick one function in your business -- marketing, sales, operations, or onboarding -- and block a half-day with the person who owns it. Map every process, every spreadsheet, every document, and every handoff point. The goal is to have the full picture of how that function actually runs before you touch any automation tool. You will find redundancies, gaps, and manual steps that no one has examined critically in years. Those are your highest-value automation targets.
Build a Centralized Strategy Dashboard
Create a single view that shows your current quarter's OKRs, the weekly actions required to hit them, and a Kanban or checklist tracking what is in progress versus complete. This does not need to be sophisticated -- a well-structured Notion or similar tool is sufficient. The discipline is in keeping it current and making sure everyone on the team references it instead of relying on ad hoc updates. Incomplete projects become visible immediately, which forces either completion or deliberate cancellation.
Define the Human-Only Role List
Write out every task each person on your team does in a given week. Then categorize each task: does it require human judgment, relationship, or creativity, or is it a repeatable, rule-based step? Tasks in the second category are candidates for automation. Tasks in the first category are where humans should concentrate. This exercise often reveals that a significant portion of paid human time is going to work that a well-configured automation could handle for a fraction of the cost.
Design an Orchestrator for One Repeatable Workflow
Choose one multi-step business process that runs regularly -- a product launch sequence, a client onboarding flow, or a content publication chain. Map every step it currently requires. Then design an orchestrator: a set of instructions (in Claude Code or a comparable tool) that can initiate and sequence the steps automatically when triggered by a single command. Build the first version narrowly, test it on one real instance, and expand from there rather than trying to automate the whole thing at once.
Establish a Shared Knowledge Base for AI Agents
Collect your brand voice guidelines, offer descriptions, standard operating procedures, and key reference documents into one accessible location. This becomes the institutional memory that your AI tools draw from. Without it, every AI interaction starts from zero context and produces inconsistent output. With it, your automations start from a foundation that reflects how your business actually operates and sounds. Even a structured Google Drive folder or a Notion wiki is a meaningful starting point.
Set Up Engagement-Based Outreach
If you have an active social presence, your existing commenters and followers are warm leads who have already signaled interest in your topic. Build or configure a system that identifies these engaged users and sends personalized outreach in your voice. Apollo and Instantly are the platforms Gray references for email; LinkedIn and X have their own engagement data. The key discipline is that the outreach must feel personal and contextually relevant, not templated, which requires the shared knowledge base from step six to be in place first.
Replace Lead Magnets with AI-Powered Tools
Gray is actively shifting from static PDFs and checklists toward interactive AI tools as his primary lead generation mechanism. The logic is that a tool delivers ongoing value and engagement, while a PDF delivers it once. If you have a core concept or framework in your business that lends itself to a calculator, assessment, or generator, building that as a Claude Code-powered tool is a higher-retention lead magnet than any document. Start with the most frequently asked question your audience has and build a tool that answers it interactively.
Full Transcript
This transcript was auto-generated and may contain errors in speaker attribution, transcription accuracy, or formatting. Long transcripts may be truncated due to processing limits. Confirm accuracy and completeness against the original source before referencing or republishing.
Matt Gray: This is how I'm going and automating 97% of my business. And I'm just going to show you behind the scenes, raw, unfiltered here, how I'm thinking. So, first things first here, this dashboard here is my project automate. Inside of this, I've gone and listed some of the core insights and strategic plans that we have across my company. Things like, you know, going and upgrading our closing team or implementing more MCP skills to go and upgrade different aspects of what we're doing, or integrating more tools instead of lead magnets, which is something I'm using Claude Code and AI for right now. Overall, in my company, we are shrinking the amount of folks that we have working in the company. We kind of are not hiring much more, and then we are defaulting to automation.
Matt Gray: So, what we have here, right, is this like, first things first, I can understand from one dashboard my strategy, right? And myself and my leadership team are all aligned around this. From there, I can see all of the weekly actions I need to go and take. Things like, you know, setting up Claude Code for different AI tools we have internally, right? Going and moving our company to a SDK, more of a shared knowledge brain across our company, being able to then also have an orchestrator which, across core processes and workflows that we have, versus even having to have someone pressing an automation and moving, say, you know, a landing page to an offer page on the website to then making email sequences. Simply having an orchestrator that we can speak to the orchestrator, which is an AI or an automation, and that orchestrator inside of Claude Code then knows, okay, if we're spinning up an offer, go and make the landing page, go and make the offer, go and make the email sequences. And then what happens is the humans on the team become more editors that are being queued up different tasks that the orchestrator is saying that we need to get done.
Matt Gray: So, inside of this dashboard as an example, I can see the weekly actions. I can see the Q2 OKRs across my entire business, right? So that we're all aligned, we know what's going on. You know, this is where the project management piece comes in. And so, there's both a calendar and a Kanban board for what we need to get done. And so, to show you this example, right? As we go and start, you know, completing different tasks, we move those into progress. As those get complete, we move them into completion. And thereby, we're not just setting these goals or just using AI to then use AI and getting lost in AI projects, right? Which I think what most companies are doing, right? There's some studies that most companies' AI investment leads to zero return. And so, what we're trying to do is make sure when we actually are building AI, we're either reducing costs and or we're increasing profits, right? That's the only reason to use it.
Matt Gray: So, we've got the calendar that shows up. We've also got a sprint checklist outlining every single piece of Founder OS in this area, being like the automation and tech side, and what are the different tasks that need to get going. So, getting managed agents in the API going, auto-generating different deliverables that we have to go and get for folks, making sure that, you know, our authorization flows are all built out. You can also see like different aspects that we're currently automating. You know, we currently are testing out something called Alberto, which will completely go and automate all of the DM flows and outreach inside of the company. On the DM side, it will handle things like LinkedIn, like X. On the cold email side, it can handle outreach using things like Instantly, Apollo to go and find folks that have gone and commented, say on my LinkedIn posts where I have 900 thousand followers over there, and then go and send personalized messages in my voice to those folks. And if they're interested, getting them set up on a call with one of my portfolio companies, depending on which one is right for them.
Matt Gray: What I like to think about is not just the growth of the revenue that AI is going to contribute to my companies, but also what automations can I put in place that will replace roles in my company. So, that if humans are on the team, they're doing stuff that only humans can do. We don't have people doing stuff that robots can do. Too many people are just vibe coding a bunch... they build all this stuff, but it's not usable. No one on their team knows what's going on. And so, I like to think when I'm building this, my process typically is that I'll take a given leader on the team, say Nick, who's my chief marketing officer, okay? And when Nick and I were making this marketing machine, I went and met up with Nick. We got a nice boardroom, big whiteboard, and Nick and I came prepared going and modeling out the whole marketing apparatus, right? What is every piece of it? What are all the documents and Google Sheets that we have across the whole company?
Matt Gray: And I wanted to make a rule for ourselves, right? Basically, let's eliminate all of these spreadsheets, and let's build a centralized place where we can understand our metrics, and where we can understand what people are working on, and hold them accountable to certain due dates. And so, similar pattern here. I have the core initiatives across all things across the marketing squad. As we, you know, check off different aspects of this, right? You can see the progress bar for that core initiative gets implemented. The cool thing is is that along with this, right? As we go and work on these things, we can start to also understand which stuff is getting behind, right? Too many founders are just letting open loops populate everywhere in their company. They have these products that are started, but then are never finished, right? And you think when you stack that up, you are just looking at the potential of your business disintegrate, right? Day by day. From there, it's also easier for the manager, in this case Nick, to see every single person on the team, right? Understand what they're responsible for this week, this month, this quarter, all with due dates, all how they're tracking as well.
Matt Gray: Now, in addition to this, this data is also coming from platforms like Asana, which is where we organize our whole content calendar. And here you can see it just synced, meaning that we go and plug in the API from Asana to this dashboard, so it's getting a live feed about every 3 minutes of anything that's changed over there, right? So, that's the core project management for our content calendar is Asana, and this is the kind of core strategy dashboard for the marketing machine. Cool thing is, you can see every single platform here, as well, LinkedIn, Instagram for me, X, and again, this is all synchronized with our Asana, which is pretty insane. All of our workshops are organized in here. You can also see the different cohorts we have going on.
Matt Gray: By the way, if you're interested in joining me over six sessions, two weeks, to go and build a beautiful cash-flowing business, I'd recommend you go and check out the link below to cashflow machine, which will be six live sessions with me, and it's going to be a lot of fun. This is actually something that I was working on with Claire, my sister, just yesterday. So, this is still a work in development, but after a 5-hour whiteboard session, which you can see the whole whiteboard over here, we came to the conclusion that there were some gaps operationally, right? And I say this just to be real, even with my businesses, right? There's constant constraints, constant problems. It's kind of just how it goes, right? As you're scaling and growing, not staying status quo, you're inevitably embarking in new areas that bring new challenges, right? So, I'm challenged all the time in my daily life, and honestly, I like it that way. So, inside of this, right? Same sort of pattern, right? It's got the command center, it's then got our whole level 10 meeting, and essentially, this is an operating system by which you go and you plan out your year, then your quarter.
Matt Gray: So, this is the ops machine, right? And this is what allows us to go and deliver on an exceptional founder experience, because this then feeds into things like our founder success machine, which helps us go and make sure that the onboarding inside of FounderOS is seamless. You know, when you go and you join FounderOS, within 24 hours, you're on a call with us, and we're doing a brand deep dive call with you. Understanding your offer, what your current status is on your marketing and positioning, where the gaps may exist, what you're really looking to do with your brand here in 2026, and we start to map out a customized roadmap for you based on exactly what stuff we need to go and build for you right now. Now, by day seven, we've gone and built your entire brand positioning for you. And for by day 21, we built your content GPS. You have all your platform strategy for every single platform that you're on. We then built out your email sequences, your newsletter, and 14 pieces of content across every platform that you're on, which then leads to the next 30 days. We're building things like your content machine and all of your content monetization as well.
Matt Gray: So, we're making sure that you have things like your newsletter, your offer stack, your lead magnets, so that we're going and taking people from your audience across social to your owned audience in your newsletter, and then driving people to your product or service or whatever it may be that you're selling. And so, that is how I'm going and automating 97% of my business. This is going to be an ongoing journey. I'll keep you in the loop. There's a lot of exciting stuff coming. I'm not done yet. Every single day chipping away on this stuff, and I'm going to keep you abreast as the newest tech that I see, so you can have more peace, profit, and purpose in life, and have a ton of fun along the way. So, if you enjoyed this video, you would love Founder OS Velocity. If you're interested, go and check out this video over here where I go deep into how I help you build a social media machine and actually do it for you. So, you can go check out that video. Be sure to go and like and subscribe, and I'll see you in the next one.
AI Master Prompt
The AI prompt on this page is auto-generated from the transcript content and is intended to support further exploration of the topics, concepts, and conclusions discussed. It is provided for informational purposes only. The user is solely responsible for all outcomes resulting from its use.
You are a strategic advisor helping me apply a business automation framework developed by Matt Gray, a founder who has built systems to automate roughly 97% of his business operations using AI, Claude Code, orchestrators, and centralized dashboards. The core of his approach is this: automation is not a technology project, it is a business design decision. Every automation must pass a two-filter test -- it either reduces a cost or increases a profit. Anything that fails that test does not get built, regardless of how technically impressive it is. The larger strategic move is to stop defaulting to human hires when capacity is needed, and instead default to automation first, with humans reserved for tasks that require genuine judgment, relationship, or creativity. The operating system he has built includes: -- A centralized strategy dashboard showing Q2 OKRs, weekly actions, Kanban progress, and sprint checklists, keeping the leadership team aligned without constant meetings. -- An orchestrator model using Claude Code, where a single command like "spin up a new offer" triggers the AI to create landing pages, offer pages, and email sequences autonomously, with humans acting as editors reviewing the output. -- A shared knowledge brain (SDK) that gives all AI agents consistent institutional context -- brand voice, offer details, operating logic -- so every automated process starts from the same foundation. -- Alberto, an engagement-based outreach system that identifies people who have already interacted with his content and sends personalized DMs and cold emails in his voice across LinkedIn, X, and email platforms. -- A whiteboard-first build process where any new automation is preceded by a full mapping of the existing department's workflows, documents, and handoffs with the relevant leader before any tool is touched. Key principles to work with: - Automation-first hiring freeze: stop replacing people with people; replace repeatable work with systems - Two-filter test: cost reduction or profit increase -- otherwise do not build - Orchestrators over doers: AI initiates and sequences; humans edit and approve - Shared knowledge brain: institutional context centralized so AI agents draw from it consistently - Whiteboard before build: map the real workflow before touching any tool - Humans do human work: judgment, relationships, creativity only -- everything else is automatable - Engagement-based outreach: warm leads from existing audience are the highest-value outreach target This is not general productivity advice, and it is not about using AI tools for personal efficiency. The framework is specifically about redesigning the organizational structure of a business so that AI and automation handle the operational layer, and humans concentrate their time on the irreplaceable parts of the work. The measure of success is business outcome (cost down, revenue up), not sophistication of the technology stack. How to use this chat: 1. Audit mode -- walk me through the two-filter test applied to a specific tool, role, or process I describe and help me evaluate whether it passes or should be cut. 2. Whiteboard session mode -- help me map out a specific department or workflow in my business, identify the manual and repeatable steps, and surface the highest-priority automation targets. 3. Orchestrator design mode -- help me design an orchestrator for a specific repeatable workflow, defining the trigger, the sequence of steps, and where human review should occur. 4. Role redesign mode -- help me analyze a current team role and identify which parts of it could be automated, what the human-only remainder looks like, and whether that remainder justifies the role. 5. Dashboard design mode -- help me design a strategy dashboard structure that surfaces OKRs, weekly actions, and progress tracking for a specific team or function. 6. Challenge mode -- push back when I am describing an automation project that sounds impressive but cannot pass the two-filter test, or when I am building complexity for its own sake. Tone: direct, systems-level, no hype, treat every suggestion as a hypothesis to be tested against real business outcomes. [Drop your current situation here -- describe the business, the team, the specific process or role you want to examine, and what outcome you are trying to drive.]