Agentic Marketing Workforce
& Operations System
We build an automated marketing team of 9 AI agents, all running on one shared Central Knowledge Base — they pass work to each other across functions, from market research to content, ad campaigns, and results summarized into LINE. Your people only press approve.
POWERED BY
Designed, built, trained, and supported by the RADA Richness strategy team · Powered by the AIYA AI agent platform
Make marketing a system, not a set of people
We design, build, and hand over an automated marketing team (Agentic AI Office) and an ad operations system that runs from campaign setup through to reporting. It solves 3 problems: a headcount bottleneck, knowledge walking out with staff who resign, and ad budget leaking unnoticed.
Built for companies and growing businesses with ad spend from THB 500,000 per month or more or THB 2–10 million in monthly sales
3 problems eating your marketing margin
Content cannot keep up with ad fatigue
You cannot produce new ad angles fast enough for audience fatigue. Cost per result climbs and sales fall with no clear reason.
People leave, knowledge goes with them
Media buyers and creatives leave often. The strategies that worked were never kept as an asset, so the system stalls every time someone changes.
High agency fees
You pay a percentage of ad spend every month, yet reports arrive late and campaigns are not adjusted in time.
What your company gets back
A system your company owns
Knowledge and strategy stay in your company's system for good, and do not leave with the employee who resigns
Creative output many times faster
Writes dozens of scripts, captions, and ad angles a day, cutting turnaround from weeks to hours
Catches wasted ad spend 24 hours a day
Catches underperforming ads and analyzes ROAS / CPA in real time, outside office hours too
Just 1 central brain
connecting all 9 agent roles
Every agent works on the same data from the Central Knowledge Base, so answers never contradict each other and work passes across functions automatically.
The full system diagram
From the moment your people give an instruction, through the central brain and the AI agent team, out to the platforms and systems you already use, and back as an asset your company owns.
Scroll sideways to see the full diagram on a small screen
Automatic handoff flow across functions
Your people only press approve at the points defined in the Human-in-the-Loop Process Map.
The 9 AI agent roles we deliver
Each role is trained on your business data, your rules, and your Brand Voice — not a general assistant that answers anything.
| Function / Team | # | Agent name | Key responsibilities |
|---|---|---|---|
| Marketing CommunicationMARCOM TEAM | 1 | Market Research Agent | Researches market trends, tracks viral topics, and gathers competitor intelligence |
| 2 | Market Strategy Agent | Reads market direction, sets campaign strategy, and turns ideas into an Action Plan | |
| 3 | Content Creator Agent | Writes articles, video scripts, captions, and graphic concepts in your Brand Voice | |
| 4 | Social Media Admin Agent | Schedules publishing, sets post times, and manages the social media content queue | |
| 5 | Report & Analytics Agent | Analyzes Engagement and Reach, and summarizes how content performed | |
| Ad operationsMARKETING PERFORMANCE | 6 | Ads Spy & Angle Agent | Decodes competitors' Winning Ads and develops new Creative Hooks |
| 7 | Ads Operation Agent | Helps set up campaigns, checks campaign status, and keeps spend under control | |
| 8 | Performance Analytics Agent | Analyzes ROAS, CPA, and Conversion Rate, and alerts you when an ad drops below target | |
| Customer RelationsCUSTOMER RELATIONS | 9 | Chat Support Agent | Answers customer chats around the clock, filters out repeat questions, and hands over cases that need a person with the full context attached. |
Scope of work we deliver
The business goal is to cut hidden headcount costs, stop ad budget leaking, and build knowledge that stays with your company, not with the employee who leaves.
Discovery & Business Learning
We interview your team, collect Business Logic, Brand Guidelines, SOPs, product data, and Customer Personas, and produce the Human-in-the-Loop Process Map
System & Workflow Architecture
We design the Central Knowledge Base structure, the Agent Memory Schema, the Guardrails, and the automatic handoff flow across functions
Knowledge Base & Model Tuning
We install semantic search and document-grounded answering (Vector Database + RAG) so answers never drift from the facts, then write the System Prompt and output standards for all 9 roles.
Integrations & Multi-Channel
We connect the Facebook Page API, LINE OA, and Ads Accounts, then install the personal AI secretary on LINE with Role-Based Access Control.
Testing, Training & Handover
We run full system testing and formal acceptance with you (UAT — User Acceptance Test), plus a 6-hour training workshop across 1 full day.
30 days of standby support (30-Day Hyper-Care)
RADA Richness specialists stay on standby and tune the prompts through the first 30 days after go-live, until the system runs stable.
The 5 delivery phases
Every phase has a concrete deliverable you accept step by step. You never wait until the end of the project to see results.
-
PHASE 1
Study the business context
About 3 working days- A process discovery workshop and interviews with the Marcom and Marketing Performance leads, to understand pain points and the targets you need to measure
- We pinpoint bottlenecks and repetitive work to convert into Automation Tasks
- We collect SOPs, policies, product manuals, Customer Personas, and Brand Guidelines
Deliverable: Business Requirement Document (BRD) and Human-in-the-Loop Process Map
-
PHASE 2
Design the system and the work-handoff map
- We define the role, decision authority, and Tone of Voice for all 9 agents
- We structure the Central Knowledge Base and the Memory Schema so every agent pulls the same data accurately
- We design how work passes across functions — the market research agent sends a brief to the content agent, which passes it on to the ads agent.
Deliverable: System architecture blueprint (System Architecture & Agent Interaction Diagram)
-
PHASE 3
Build the knowledge base and tune models
- We load your documents and business rules into the system, turn them into a Knowledge Base, and set the AI to pull answers from your real documents before it answers (RAG).
- We write the System Prompt and role-specific rules, such as content formulas and campaign setup criteria
- We set the Guardrails the AI cannot cross, so it never answers outside its scope, and set report, caption, and brief templates to your company standard.
Deliverable: A working Central Knowledge Base and Agent Modules with initial training complete
-
PHASE 4
Connect systems and set up channels
Add-on Scope- Connect the Facebook Page API and LINE Official Account for the Marcom team
- Connect Ads Accounts (Facebook Ads / Google Ads) for the Marketing Performance team
- We install a personal AI secretary that links each employee's LINE account to the Central Hub, with access rights set per person.
Deliverable: An API Integration layer ready to exchange data with external platforms, plus a personal AI secretary for each employee
-
PHASE 5 · GO-LIVE
Test, train, and hand over
- We test data accuracy and simulate real situations (Stress & Edge Case Testing), with formal acceptance alongside your team
- We run a workshop on prompting, reviewing AI output, and using the secretary on LINE, and hand over the user manual.
- Go live, with monitoring and accuracy tuning through the 30-Day Hyper-Care period
Deliverable: A live system, a user manual, and 30 days of continued support
A 6-hour workshop makes your team the operators
Intensive AI Marketing Team Workshop for your marketing communication team, ad operations team, and the executives involved
Morning — foundations and using the Marcom AI Team
Afternoon — ad operations, coaching, and real cases
- Slides and training handouts
- Standard Marketing Prompt Cheat Sheet & Templates
- Recorded video of the training for later review
Hire every year, or invest once
A straight comparison between adding 3–4 headcount and investing once in an Agentic AI Marketing Team
Add 3–4 headcount
- ✕Media buyers and creatives leave often, and what your team learned about Winning Ads walks out with them
- ✕Real cost runs above salary — social security, bonus, benefits, equipment, software
- ✕Interviewing and retraining costs you time every time someone changes
- ✕1–2 pieces of content a day, ads checked only during working hours
- ✕Human error in budget changes or audience selection burns ad money for nothing
AGENTIC AI MARKETING TEAM
- ✓You get a system that is a company asset — knowledge and strategy stay in the Central Knowledge Base permanently
- ✓One investment, with no per-head costs rising every year, such as bonuses, benefits, and training new hires
- ✓Writes dozens of script and caption variants in one pass
- ✓Ads are monitored 24 hours a day, so less budget burns outside office hours
- ✓Decisions run on data, not on gut feel
Accuracy and standards you control
AI delivers work in your Brand Voice and Business Logic every time, so you do not review the same thing twice
Your people become the operators
Nobody is laid off. Routine work is cut, and your people move into strategist and AI operator roles that are measured on business results.
Handoffs that never stall
The system pulls data across Research, Content, and Ads, and reports into LINE immediately
9 hires versus one Agentic system
To cover all 9 functions with people, how many roles would you hire and what would it cost — against a one-time system investment?
| With human hires | With AI staff | |||
|---|---|---|---|---|
| # | Role to hire | Market salary (THB/month) | Replaced by AI staff | AI staff fee (THB/month) |
| 1 | Market ResearcherMarket Research | 25,000 – 35,000 | Market Research Agent | 3,000 |
| 2 | Marketing StrategistMarket Strategy | 40,000 – 60,000 | Market Strategy Agent | 3,000 |
| 3 | Creative / ContentContent Creator | 25,000 – 35,000 | Content Creator Agent | 3,000 |
| 4 | Social Media AdminSocial Media Admin | 18,000 – 25,000 | Social Media Admin Agent | 3,000 |
| 5 | Content AnalystReport & Analytics | 30,000 – 40,000 | Report & Analytics Agent | 3,000 |
| 6 | Ad CreativeAds Spy & Angle | 30,000 – 40,000 | Ads Spy & Angle Agent | 3,000 |
| 7 | Media BuyerAds Operation | 35,000 – 50,000 | Ads Operation Agent | 3,000 |
| 8 | Ad Performance AnalystPerformance Analytics | 35,000 – 50,000 | Performance Analytics Agent | 3,000 |
| 9 | Chat SupportChat Support | 20,000 – 28,000 | Chat Support Agent | 3,000 |
| All 9 roles | 258,000 – 363,000 | All 9 AI staff roles | 27,000 | |
First-year cost
12 months of salary plus around 30% in hidden costs — social security, bonus, benefits, equipment, software, and recruitment
- ✕A cost that repeats every year and rises with salaries
- ✕Works office hours only, takes sick leave and holidays, and can resign
- ✕Knowledge sits with people, not with the company
First-year cost
System setup THB 550,000 + monthly AI staff fee THB 27,000 x 12 months — add-on fees for the channels and systems you actually connect are not included.
- ✓From year two you pay only the monthly AI staff fee, with no setup fee again
- ✓Works 24 hours a day, no leave, no resignations
- ✓Knowledge and strategy stay in your Central Knowledge Base permanently
Note: salary ranges reference Bangkok market rates for mid-level roles · hidden costs estimated at about 30% of salary · these figures compare cost structures and are not a guarantee of returns · the Agentic system is designed to cut routine work, not to replace your entire staff
2 main fees, plus add-ons
We break the fees out so you see what you pay for, instead of one lump sum. There are 2 main parts: a one-time setup fee and a monthly AI staff fee. Platform connections and external API integrations are add-ons, charged only for what you actually use.
Core Setup
Main system setup, one-time flat fee
- ✓Central Knowledge Base architecture
- ✓Build and train all 9 AI agents
- ✓Security and the rules the AI cannot cross (Guardrails)
- ✓1-day training + 30 days on standby
- ✓20 user accounts included (add more, see add-on fees)
Monthly AI staff fee
Priced by the number of agents you switch on, from THB 3,000 each
- ✓Ongoing system and integration maintenance
- ✓Standard quota of 10,000 Token per agent per month (a Token is the unit that counts the text an AI processes) — top up if you go over quota, see add-on fees
- ✓Continuous accuracy tuning for every agent
A personal AI secretary on LINE for every employee — give instructions, look up information, and receive reports in the chat app your team already uses every day, connected straight into the Central Hub
Add-on fees
Charged only for what you actually use, separate from the 2 main fees above — platform connections, external API integration, extra user accounts, and Token top-ups
Platform Connect Fee
Platform Connect Fee · priced by the number of channels you switch on
- ✓Facebook Page
- ✓LINE OA
- ✓Facebook Ads
- ✓Google Ads
Extra user accounts
User Account Fee · for admins added to the back office beyond the 20 accounts already included in the setup fee
- ✓Access to the Agentic back office
- ✓Brief and review agents within your permissions
- ✓Set data access rights per person
The personal AI secretary on LINE for every employee stays inside the original scope, at no extra cost
External system integration
System Integration · priced by team working days (Man-day)
Minimum 3 Man-day = THB 54,000
- ✓API connection to your existing CRM
- ✓API connection to your existing ERP
- ✓Any other internal system with an open API
The actual number of Man-days is estimated from system complexity and stated in your quotation
Token Top-up
Token Top-up · for months when you go over the standard quota
- ✓Use it when the standard 10,000 Token per agent per month is not enough
- ✓Charged per top-up, separate from the monthly fee
- ✓In months you stay inside the quota, you pay nothing here
The Token volume per top-up is stated in your quotation
Questions executives ask
The questions executives actually ask us — from how this differs from ChatGPT to how your data is kept safe
Still no answer? Ask usHow is Agentic AI different from letting the team use ChatGPT?
ChatGPT is a separate chat box. Every new session starts from zero context, and nothing is handed on to the next task. Here, all 9 agents read from the same Central Knowledge Base, remember your business context, and pass work to each other across functions along a designed flow. Your people step in only where an approval is required.
Do we have to let marketing staff go?
No. The system is built to cut repetitive routine work and move your existing people up to running it — as strategists and AI operators who brief, review, and measure business results. That is why we include a full 1-day workshop in the scope of work.
Could the AI give a wrong price or promotion and cause problems?
This is exactly why we use a Vector Database and RAG — agents answer from your real documents, not from guesswork — and why Guardrails define which topics they must never answer alone and must pass to a person. We run a Stress Test & Edge Case Simulation before every go-live.
How safe is the company data we put in?
The knowledge base belongs to your company, and Role-Based Access Control sets exactly who can see which set of data. We put this in place in Phase 2, before any real data enters the system.
How big does a business need to be for this to pay off?
It fits companies spending from THB 500,000 per month on ads, or with THB 2–10 million in monthly sales. The payback comes from two directions: the headcount you no longer add, and the ad budget you stop wasting. If your business is smaller than that, start with Performance Marketing or Marketing Communication first.
Who actually does the work, RADA or AIYA?
Your contracting party, responsible for all the work, is RADA Richness Co., Ltd. — strategy, system design, training, and support after handover. AIYA is the AI agent platform we use as the engine. You deal with, and sign off with, the RADA team only.