Human First Media
The system for coaches, consultants, and business owners who want AI agents that know their voice, run their operations, and stop needing constant hand-holding.
By Meredith Medland · Human First Media
The Real Problem
You have tried the prompts. You have pasted your bio into ChatGPT and asked it to write a post. The result sounds like a press release from someone who skimmed your LinkedIn for 30 seconds. Generic, hollow, and nothing like the way you actually talk.
The problem is not the technology. The problem is that AI does not know you. It has no context about your positioning, your offers, or the specific people you serve. Every time you open a new chat, you start from zero.
There is a better way to work with AI. One where your AI team carries your voice, understands your business, and operates with enough context to produce work you can actually use.
This is the system that makes that real.
What You're Building
You are not building one chatbot. You are building a team of AI agents, each with a defined role, a clear job description, and a set of tasks they execute on a schedule. And you are building an AI Co-Founder who oversees all of them.
Your AI Co-Founder creates new agents when your business needs them. It writes their job descriptions and skill files. It runs weekly performance reviews, scores each agent on output quality and brand voice alignment, and reports back to you with a clear picture of what is working.
Your tools connect to Claude (email, Slack, calendar, CRM, Google Drive). Claude reads what is actually happening in your business before generating anything. That context is what separates this from every other AI workflow you have tried.
Your connected tools give Claude real data: emails, conversations, calendar patterns, recurring tasks.
Based on what it finds, Claude surfaces the tasks your team is doing manually that AI should handle.
It creates the agents you need, names them, writes their job descriptions and skill files.
Daily, weekly, and monthly workflows fire automatically. Humans review and approve the output.
Your AI Co-Founder scores every agent, flags what needs attention, and tells you where to optimize.
Foundation First
Here is what most people skip. They connect tools and start building agents before their AI knows anything about who they are or how they talk to clients. Then everything sounds generic because it is.
Before you hire a single AI agent, install your Foundational Brand Skills. These are five files that teach Claude your business at a structural level.
These five skill files are the foundation everything else runs on. Skip them and your agents will produce work that sounds like anyone. Install them and your agents will sound like you from the first draft.
Before You Start
Claude.ai with Cowork access. This is where your AI team lives and operates.
Connect at least one tool you use daily. Gmail, Slack, or your CRM is the best starting point.
Five foundational skill files that teach Claude who you are. Install them before building anything else.
The full system gets built in a single session. After that, it runs on its own schedule.
The Build
Before Claude can tell you what AI should handle, it needs to see what is actually happening. Connect the tools where your biggest time drains live: email, team chat, CRM, calendar. Then ask Claude to find the patterns you have stopped seeing because you are too close to them.
Read through my [email / Slack / CRM] from the last 30 days. Give me a list of recurring tasks, bottlenecks, and manual work you see happening. Organize it by what happens daily, weekly, and monthly.
Once Claude has read your tools, have it turn the findings into a full operating rhythm. This becomes the blueprint for your AI team.
Based on what you found, what AI tasks could we run daily, weekly, and monthly to streamline and scale this business? Be specific about who on the team is currently doing each task and how long it takes them.
Your human team is doing the AI's job: writing first drafts, manually routing emails, chasing payment follow-ups, copying data between tools. These tasks are predictable and repeatable, which makes them ideal for AI agents.
Take the task list and ask Claude to design your AI team. Each role should own a clear slice of your operations with no overlap. Give your AI agents real names. It changes how you and your human team relate to them. "Ask Nova to pull the research" lands differently than "run the AI research prompt."
If you were going to create AI team members to handle these tasks, who would they be? Give each one a first name, a full title, a one-sentence job summary, and their primary tasks (daily, weekly, monthly). The goal is zero overlap between roles.
Before you build anyone else, build the one who manages everyone. Your AI Co-Founder is the team lead. It oversees hiring, onboarding, performance tracking, and weekly scoring for every other agent on your team. This is the step most people skip. They build a collection of individual agents and then wonder why nothing coordinates. The AI Co-Founder is what turns a collection into a team.
Create the full job description for my AI Co-Founder. Include: their name and title, their personality and communication style, what they do daily and weekly, what they never do without my approval. They are responsible for overseeing all other AI team members.
Take everything from Step 4 and turn it into the actual skill file. This is the file Claude reads every time your AI Co-Founder activates. Include their scheduled tasks directly in the skill file, because that is what makes them proactive instead of reactive.
Take everything we've built for my AI Co-Founder and write their complete SKILL.md file. Include the YAML frontmatter, personality section, all responsibilities, all their scheduled tasks with step-by-step instructions, and their weekly performance review cadence with scoring rubric.
Create the file at /mnt/skills/user/[ai-co-founder-name]/SKILL.md in Cowork. Paste the full skill content. Save. Open a new session and test. Your AI Co-Founder needs to be live before they can build anyone else.
Here is where the system starts working for you. You do not write the job descriptions for the rest of the team. Your AI Co-Founder does. Pull them up, give them the task list from Step 2 and the team structure from Step 3, and let them run.
You are now managing the build of our AI team. Write the full job description for [Name]. Include: who they are, their personality, every task they own (daily, weekly, monthly), how their output is reviewed, and what they never do without human approval.
Your AI Co-Founder writes the SKILL.md files too. For each agent, Claude builds the file. You copy it. You create the file in Cowork. That agent is now hired.
Take everything we've built for [Name] and write their complete SKILL.md file. Include the YAML frontmatter, personality, all tasks, scheduled cadences, and approval rules.
For each agent, create their file at /mnt/skills/user/[team-member-name]/SKILL.md. Paste. Save. Test each one before moving to the next. Do not batch-install without testing.
Every agent needs a clear schedule. Daily tasks run every morning. Weekly tasks run on a set day. Monthly tasks have a deadline and a delivery destination. Every automated task needs somewhere to land: a Slack channel, a Google Doc folder, a tagged email draft.
For [Name], write out their complete task schedule. Include: what they do every day and at what time, what runs weekly and on which day, what runs monthly, and where each output is delivered.
Once your AI team is running, you manage them with a five-minute weekly check-in. Your AI Co-Founder scores every agent, flags what needs attention, and gives you a clear picture of team performance.
Run the weekly AI team check-in. For each team member, tell me: their status (active / needs attention), what they produced this week, their score on output quality and brand voice alignment, and one recommendation.
Week one, your AI team is functional. Week four, the agents are calibrated to your voice and patterns. By week twelve, your AI team knows your business well enough that first drafts require minimal editing. The weekly check-in is what makes that trajectory possible.
The Outcome
Five foundational brand skill files that teach every AI agent your voice, your clients, your offers, and your positioning from day one.
A complete AI team org chart with named members, clear roles, and no overlap between jobs.
An installed AI Co-Founder that manages the whole team, runs weekly performance reports, and scores every agent.
A SKILL.md file for every agent that lives permanently in Cowork and loads fresh every session.
A daily operating rhythm where AI generates first drafts and humans review and approve.
A weekly performance review with agent scores that tells you exactly what is working and what needs adjustment.
A system you can give to clients so they can build their own AI team using the same structure.
Quick Reference
Copy these into Claude in sequence. Adjust the bracketed sections for your business. Install your brand foundation before running Prompt 1.
Install your five foundational brand skills (Identity, Voice, Ideal Client, Offers, Market) before building anything else.
Read through my [email / Slack / CRM] from the last 30 days. Give me a list of recurring tasks, bottlenecks, and manual work. Organize by daily, weekly, and monthly.
Based on what you found, what AI tasks could we run daily, weekly, and monthly to streamline and scale this business? Be specific about who is doing each task now.
Create AI team members to handle these tasks. Give each a first name, title, one-sentence job summary, and their primary tasks. Zero overlap between roles.
Create the full job description for my AI Co-Founder. Include name, title, personality, daily and weekly responsibilities, and what they never do without my approval.
Write the complete SKILL.md for my AI Co-Founder. Include YAML frontmatter, personality, all responsibilities, scheduled tasks with instructions, and the weekly scoring rubric.
Create the file at /mnt/skills/user/[ai-co-founder-name]/SKILL.md in Cowork. Paste. Save. Test before continuing.
You are now managing the build of our AI team. Write the full job description for [Name]. Include personality, every task, review process, and approval boundaries.
Write the complete SKILL.md for [Name]. Include YAML frontmatter, personality, all tasks, scheduled cadences, and approval rules.
For each agent: create /mnt/skills/user/[name]/SKILL.md. Paste. Save. Test before moving on.
For [Name], write their complete task schedule. Include daily timing, weekly cadence, monthly deliverables, and where each output is delivered.
Run the weekly AI team check-in. For each team member: status, what they produced, their quality and brand voice score, and one recommendation.
Build and install your AI Co-Founder using Prompt 5 before you build anyone else. Then have your AI Co-Founder write the job descriptions and skill files for the rest of the team. That is the sequence that makes everything coordinate.
This system is running inside Human First Media right now. The same structure is available to every client. One focused hour, nine prompts, a team that compounds week over week.
Learn More