
One-Person Agency Marketing: Using AI Agents to Handle Repetitive Work
One-person agency marketing is a model where one person manages many advertising clients, using AI agents to handle structured and repetitive work every week. Most of an advertising agency's time is not spent on ideas but on reports, account checks, and tracking spending progress. According to Yarchi (X, 18/10/2026), these three tasks are exactly the kind of work an AI agent does best. This post guides you on how to hand them over to an agent like Claude Code from Anthropic, while you keep only the decision-making part. For example, a weekly report that takes a whole day can become a draft waiting for you to read at 8 AM on Friday.
This post is based on the process "How to build a one-person marketing agency with AI" by Yarchi (@undefinedKi), rewritten in Vietnamese by OPCAIOS, adjusted with examples for the Vietnamese context, and focused on how to build it using available tools. The original post had sponsorship for a paid AI product; that advertising part is not included here.
What Are the Stages in an Agency's Loop?
An agency's loop has five stages that every client goes through, repeated over and over. According to Yarchi (2026), they are:
- Close the client: learn about prospects, prepare a pitch.
- Audit: find what's broken in the advertising account.
- Run: keep spending on track, every day.
- Report: show the client what happened and next steps.
- Scale: turn what works into a brief for the next campaign.
Closing and scaling need analysis and humans. The three middle stages—audit, run, report—are mostly about gathering numbers, checking against rules, and writing documents. These three stages take up nearly your whole week, so this is where an agent helps most. The principle throughout: agent collects, audits, and drafts. You decide.
Five Layers of a Working System
If you're missing one layer, the system breaks.
1. Access: What can the agent read?
Access is the list of data sources the agent is allowed to read: advertising accounts (Meta, Google, TikTok), Google Analytics, your store (Shopify, Haravan, Sapo), and Google Sheets. Start with read-only access, since the agent doesn't need to write anything yet. Next, the simplest way is one MCP connector per platform. For example, a read-only connector for a Meta Ads account is enough for the agent to get daily spending without being able to change the budget. See the post learning MCP free with Hugging Face if you're new to MCP.
2. Agent: Who does the work
You need an agent that can work with files and code, not just chat—for example Claude Code or Codex running in a project folder:
agency/
├── clients/
│ └── client-a/
│ ├── rules.md # goals, limits, when to report
│ └── reports/ # everything the agent produces
├── skills/
│ ├── audit/SKILL.md
│ ├── spending-pace/SKILL.md
│ ├── weekly-report/SKILL.md
│ └── creative-brief/SKILL.md
└── log/runs.csv # one line per run
3. Memory: What does the agent know about each client?
Each client's rules.md file is what affects the system's quality the most. Every time you fix something by hand from the agent, add a line here so it doesn't repeat next time. For example:
Target metric: CPA, goal 150,000đ
Budget: 100 million/month, Google + Meta
Never: turn on paused campaigns, touch brand campaigns
Always: note the source and date range next to each number
Report right away: spending pace off by 10%, any change over 5 million
Report voice: simple words, numbers first
(The numbers above are just examples.)
4. Clock: When does work start?
Automatic schedule (cron, scheduled task) so the agent starts by itself without you asking: 7 AM every day, 8 AM Friday, first Monday of the month.
5. Review gate: What needs you?
End every skill with one line: post the draft and wait for approval, never apply changes on your own. This is the layer that lets you sleep well.
Start with one client and one process. Only add the next process when the first one runs smoothly without needing you to step in.
One template for every process
Everything in an agency follows the same seven-step structure. Once you understand it, building a new process takes just a few minutes:
- Activate: on schedule, when a new client arrives, when a message comes in.
- Fetch data: new numbers from connected accounts.
- Cross-check: against rules in the client's
rules.md. - Create output: a file (spreadsheet, PDF), not a chat box answer.
- Review: you approve or you don't.
- Send: to the client or into an ad account.
- Log: one line, so the next run knows what the last run did.
Write this template once as a skill, then copy it for each process below:
---
name: [process-name]
description: [what it creates]. Use when [activation condition].
---
1. Read clients/[client]/rules.md before everything else.
2. Fetch [data] within [time range] from all connected accounts.
3. If one source fails, stop and say which one. Never guess numbers to fill the gap.
4. Cross-check the numbers against the rules: targets, limits, when to report.
5. Create [output]. Each number with its source and time range.
6. Post a draft and a 4-line summary. Wait for approval.
Never send on your own or change anything.
7. Write one line to log/runs.csv: date, client, what you did, result.
Step 3 matters most. A process that quietly fills in guessed numbers where data is missing is worse than having no process at all.
Closing a client: how to research a prospect before the pitch?
Researching before a pitch is the agent gathering public information about a prospect into a sourced summary. Most pitches fail because of poor prep, and with an agent, prepping each prospect takes just a few minutes. The process has four parts:
- Activate: a new prospect appears.
- Input: their website, ads they are running, two or three competitors.
- Output: a sourced summary and an outline for a pitch deck.
- You decide: whether to pitch or not, and which point to open with.
Public data is enough for most of this: Meta's Ad Library and Google's Ads Transparency Center show every ad a business is running. Here is a sample prompt:
Research [company] before the pitch.
1. What they sell, to whom, at what price.
2. Ads running on Meta and Google: channels, angles, formats,
how long the longest-running ad has been live.
3. Two competitors: where they position themselves differently.
4. Three gaps we can fix, each with evidence and a link.
5. An outline for a 6-slide pitch deck around those three gaps.
Mark each claim as "verified" or "guess". The outline cannot use guesses.
The last line is the most important one: a pitch built on a wrong guess about the client's own business ends very quickly.
Audit: how do I find what is broken in an ad account?
An audit is the agent comparing real numbers against four categories to find where money is wasted or measurement is wrong. Run it when you sign a new client, then once a month after that. According to Yarchi (2026), a good audit is also the most persuasive argument when pitching. The process has four parts: activate when a new client signs on, then monthly; input is the ad account, analytics, store data, and search keyword reports; output is a spreadsheet with a health score and one fix for each area; you decide what to fix this week.
Four categories cover most ad spend:
- Spend pace: is each channel spending on track for the month.
- Wasted keywords: keywords burning money but not closing deals.
- Measurement gaps: conversions counted twice or not counted at all.
- Ad fatigue: your best ads are losing power.
Audit [client] within [time range].
For each area (spend pace, wasted keywords, measurement, ad fatigue):
status: Fix now (this week), Review (this month) or OK
score from 0 to 100, main finding, estimated monthly impact
evidence: which report, which numbers
one fix proposal
Add total health score and total monthly savings potential.
Export as spreadsheet: overview tab, evidence tab.
Propose only. Do not change anything in the account.
Operations: How to track daily spending pace?
Spending pace is the ratio between actual spend and planned spend up to today, and the agent checks this ratio every morning for every customer and every channel. The process has four parts: triggered at 7 AM each day; inputs are channel spending, monthly budget, and goals in rules.md; outputs are a spending pace table with one suggestion for each customer running off pace; you review or reject each suggestion.
The logic has four steps:
- Pace = actual spend ÷ planned spend up to today.
- Off pace within ±10% is on track; outside that range is overspend or underspend.
- Each off-pace case: one suggested change, with reason and numbers.
- Nothing off pace means no message, so you only hear about customers who need you.
Hypothetical example: 30 million đồng budget per month, today is day 10 of a 30-day month, so planned spend up to today is 10 million đồng. If actual spend is 12 million đồng, pace is 120%, off by 20%, and the agent suggests a cut.
7:00 each day, for each customer in clients/:
get spend since start of month by channel
pace = spend ÷ (budget × current day ÷ days in month)
flag every channel off pace outside ±10%
for each flagged channel, suggest one budget change: from, to, reason, expected pace
Only send flagged cases. Don't change anything until I approve.
Upgrade step: build an internal tool once. Ask the agent to build a small web app the team opens each morning, with customer list, pace indicators, spending chart by day, and approve buttons. According to Yarchi (2026), with a coding agent this is an afternoon's work, running on a laptop or a cheap VPS.
Reports: what goes into a weekly customer report?
The weekly report is a 3-page document the agent drafts from ad platform and store data, which you review and then send to the customer. The report is what the customer actually sees from your work, and many agencies spend a whole day on it. The process runs at 8 AM Friday, takes this week's and last week's data, outputs a 3-page PDF with a 4-line summary sent to the customer's chat group like Zalo or Slack.
Each report answers three questions:
- What happened: one headline in plain language, key metrics versus last week.
- Why: daily trends and the ads behind them.
- What next: actions, flagging what needs customer approval.
Rule to avoid disputes: platforms like Meta or Google often log more conversions than actual store orders, so take revenue and order count from the store and source every other number.
8:00 Friday each week, make weekly report for [customer], [Monday to Sunday last week].
Page 1: one-sentence headline, spend, ROAS, CPA versus last week.
Page 2: charts of spend and ROAS by day, top 3 ads with results.
Page 3: what changed, why, next steps. Flag items needing customer approval.
Revenue and order count from store. Source all other numbers clearly.
Export branded PDF and 4-line summary. Post as draft for me to review.
Growth: the creative feedback loop
Customers stay when results keep improving. Results improve when each new ad builds on what already works.
- Triggered: once a month.
- Input: every running ad and results from the last 30 days.
- Output: creative brief based on common patterns in winning ads.
- You decide: which tests to run.
The loop has five steps: get results for every ad → find what winning ads share (format, opening line, length, angle, voice) that the bottom 50% don't → write brief → test each variable against the best current ad → if the new ad beats the control by 10%, scale the budget; if not, pause and note why.
Once a month, for [customer]:
Get every ad from the last 30 days with spend, ROAS, and CPA.
Find what the top 20% share that the bottom 50% don't:
format, opening line, length, angle, voice.
Write next brief:
hypothesis with expected result, 3 opening lines, scenes to shoot,
format, and one test against the current best ad.
Post as draft. I choose which test to run.
What a Week in a One-Person Agency Looks Like
A week in a one-person agency is a work week where five processes have already been scheduled, so you just review and decide. According to Yarchi (2026), the template schedule has four checkpoints. Each morning, the performance metrics for every client are ready before you turn on your computer. For each new prospect, the discovery is completed before the first call. The first Monday of the month, the account check and creative brief are written. Finally, 8 AM Friday, every weekly report is written and waiting for you to read.
Your part of the work comes down to three things: reviewing changes, talking with clients, and deciding the next step. For example, with 5 clients, you read 5 draft reports on Friday morning instead of writing each one yourself. The work still gets done, it's just that you're no longer the one who has to start it.
Work You Never Fully Hand Off to an Agent
Two questions decide who keeps a task: can it be undone, and does it need judgment?
| No judgment needed | Judgment needed | |
|---|---|---|
| Can be undone | Get data, write draft, check, discover — agent does it themselves | Pitch angle, pick test — agent suggests options, you choose |
| Cannot be undone | Change budget, send report — agent proposes, you review | Service pricing, contract, take or drop client — only you |
Gradually give agents more autonomy:
- Week 1: only write drafts.
- Weeks 2 to 4: take action after you review.
- From month 2 onwards: make small changes on their own within limits in
rules.md.
Any work that cannot be undone must always go through a review step.
Where Do You Start Building a One-Person Agency?
Start with one client and one process that eats the most time in your week, usually the weekly report. First, build the process by the seven-step template. Next, run it in parallel with the old way for a few weeks to compare results. Finally, only add the next process once the first one runs smoothly.
If you're working alone and want to know what to automate first, read more at What Should You Automate First If You Work Alone? and The AI System I Use to Run My Work Alone. Need someone to help you build the right process for your work, check out OPCAIOS's SOLO program.
Read more: Breaking Down How a Micro-Agency AI Operates; From "Selling Your Time" to "Selling Results".
Source: Yarchi — How to build a one-person marketing agency with AI (X, 18/10/2026), rewritten and adapted by OPCAIOS.
Translated from Vietnamese with AI. Read the original.
Want to apply this in your business?
Book a free 30-minute call (Vietnam time, in English or Vietnamese).
