Version 0.2 – 3 Jun 2025
1. Why you’re seeing this
We are building an automated pipeline that turns our existing (sometimes outdated) product‑page content into copy and images that are:
on‑brand and conversion‑optimised The automation needs structured inputs from Marketing before it can run. Everything below is written in plain language and shows an example next to every field you need to fill. If a row is blank, please complete it once for each product and return the file. (Feel free to duplicate the tables in a Google Sheet.)
2. One‑Page Master Checklist ✅
What we need (one row per product)
💡 Tip: If several products share the same brand‑voice link, just point to the same URL each time.
3. Tables to copy‑paste
Table A – Ingredient list
Source (Organic/Conventional)
Table B – Consumer segment / persona
{
"segmentName": "Millennial Women 25‑35",
"demographics": "Female • 25–35 • Tier‑1 cities",
"needState": "High stress + poor sleep",
"shoppingDrivers": "Natural, non‑habit forming"
}
4. How your inputs feed the agents
Fields Marketing must supply
5. Ready‑to‑use Templates (copy the JSON / tables)
We’ve pre‑built machine‑readable templates so you don’t have to guess the format.
Ingredient CSV – just grab the heading row from Table A. Persona JSON – copy Table B. SEO keywords – comma‑separated plain text. If you prefer Google Sheets, keep one tab per product and use the same column names; the tech team will convert it automatically.
6. Visual Agent Flow (schema diagram)
Below is a high‑level flowchart showing how the agents hand off to each other. (If you’re viewing this file in a Mermaid‑enabled tool such as GitHub, the diagram will render automatically; otherwise, follow the text arrows.)
graph TD
subgraph Inputs
A1[Existing Content] --> A
A2[Ingredient List] --> A
A3[Consumer Segment] --> A
A4[Competitor Data] --> A
end
A(Input Aggregation) --> B[Ayurvedic Science Expert]
A --> C[Modern Science Expert]
B --> D[Marketing Positioning Agent]
C --> D
D --> E[Marketing Copy Agent]
D --> F[Graphic‑Design Brief Agent]
E --> G[Medical Compliance Agent]
F --> G
E --> H[Regulatory Compliance Agent]
F --> H
%% Iteration loops – dashed lines show change‑requests
G --|Change Req|--> E
G --|Change Req|--> F
H --|Change Req|--> E
H --|Change Req|--> F
G --> I[Output Packaging]
H --> I
7. Detailed Input Reference (fill one row per product)
This expands the quick checklist into explicit fields the automation expects. Copy the headings into a Google Sheet or CSV.
Tip: You can add additional columns; the pipeline ignores unrecognised headings.
8. SOP Cheat‑Sheet for Each Agent (for visibility)
What it does (internal steps)
SOPs are coded into the pipeline – Marketing just needs to supply the inputs above.
9. Glossary (plain English)
USP – the single most compelling reason a customer should pick us over others. RTB (Reason to Believe) – evidence that backs up the USP (e.g., clinical study, Ayurvedic citation). Tone of voice – how the brand “sounds” (e.g., serious medical, warm & friendly). 10. Iteration Loop & “Edge‑vs‑Compliance” Guardrails
When either Medical (G) or Regulatory (H) Compliance issues a Change Req flag, the system automatically routes their structured feedback back to the Marketing Copy (E) and Design Brief (F) agents. The loop is governed by these rules:
How the loop works
Change‑Req payload lists offending text snippets + rule ID (e.g., FSSAI‑4.5 or DiseaseClaim‑Critical). Copy/Design agents re‑draft only the flagged sections, preserving brand voice & SEO keywords. NLP engine recalculates Brand Punch; if <75 %, copy is enriched (tone verbs, emotional language) before resubmission. On the 3rd failure, the pipeline pauses and notifies humans to decide whether to soften claims or provide more evidence.