About Platform Resources Articles Contact
IMA AI  ·  Resources

The Brand
Launch Playbook

How to take a brand from nothing to a live, AI-assisted storefront — in one repeatable order, so every launch is faster than the last.
Version v1.0
Date June 2026
Free No form. No email.
The Problem

Every new brand starts from zero.

Without a standard order, each launch reinvents the same wheel.

A new brand lands and the work gets improvised. Someone builds a page, someone else writes product notes, and the AI assistant is bolted on at the end with whatever information happens to be lying around. The result is slow, inconsistent, and different every time — and the knowledge that should power the assistant ends up scattered across chats, docs and people's heads.

We run a portfolio of brands, so we stopped improvising. This is the single path we follow to launch any brand — website to chat assistant — so the order is never guessed again.

What You Build

The brand stack.

Four pieces. The same four, every brand.
PieceWhat it isRequired?
WebsiteThe brand's story, products and SEO — the public face.Required
ShopThe online store, where the brand actually sells.Optional
Knowledge baseThe facts your chat assistant answers from.Required
Chat assistantAI that answers visitors and passes you only warm leads.Required

One team owns the public surfaces — the website, the shop and the assistant's knowledge. The live assistant itself runs on a separate AI platform. Knowing who owns what before you start is half the battle.

The Order

Collect → website → shop → KB → assistant.

Each step feeds the next. Do them in this order.
1Collect infoBrand facts, product list, positioning, policies, contacts — in one place.
2Build the websiteThe public site. Building it forces you to organise the brand's core information.
3Build the shopIf the brand sells, stand up the store. This is where prices, SKUs and stock live. (Optional.)
4Build the knowledge baseAssemble what the assistant answers from — drawn from the website, the shop and your support knowledge.
5Launch the chat assistantDefine how it behaves, load the knowledge base, set the allowed domain, and go live.
The Logic

Website first. Knowledge base after.

WhyThe website is the forcing function.
Building the site is what forces you to organise the brand's real information — products, claims, story. That structured content becomes the backbone of the knowledge base. Try to write the KB first and you're organising facts with no home to put them in.
Why notKB-first is wasted effort.
A knowledge base written before the site means doing the same work twice. It comes after the website — and after the shop, because pricing and stock live there.
This isn't theory. It's simply how it already happens once you've launched a few brands. We named the order so nobody has to rediscover it the hard way.
The Knowledge Base

Your KB is a superset.

It's more than the website text — and the third source is the one teams forget.
SourceWhat it addsExamples
The websiteYour public content.Brand story, products, ingredients, the science.
The shopThe commerce facts.Prices, SKUs, stock, shipping & returns.
Support knowledgeAnswers that never made it onto the site.FAQs, usage & troubleshooting guides, policies, the replies your support team gives daily.

KB = website + shop + support knowledge. Most real questions live in that third bucket — capture it, or your assistant will sound thin.

Where It Lives

Keep the assistant's brain in version control.

Your assistant runs on an AI platform — but its brain (how it behaves, its settings, and its knowledge files) should live as plain files in your site's code repository, versioned like everything else. The running copy is uploaded from there.

RuleThe repo is the source of truth.
Edit the behaviour, settings and knowledge as files, then push them to the live platform. Never edit only on the platform — you lose the ability to see what changed, diff it, or roll back.
Non-Negotiables

Two rules you can't skip.

Rule 1Keep the brain private.
Your assistant's instructions and knowledge must never be publicly visible on the site. Most static hosts publish every file in the repo — so a folder of knowledge files can leak straight to the public. Add a redirect that blocks that folder, and verify it after every launch.
Rule 2Customer data never goes in the repo.
Leads, chat transcripts and visitor details belong only on the assistant platform, handled under your privacy obligations (in Malaysia, the PDPA). Never commit customer data to a code repository.
Every Launch

The launch checklist.

1CollectGather brand, product, policy and contact info in one place.
2WebsiteBuild and publish the site — it organises the core information.
3ShopIf selling, stand up the store (prices, SKUs, stock).
4Knowledge baseAssemble website + shop + support knowledge into knowledge files.
5AssistantWrite its behaviour and settings, load the KB, set the allowed domain, go live.
6GuardBlock the knowledge folder from public view; confirm no customer data sits in the repo.

Built for founders and teams launching brands in Malaysia who want every launch faster and more consistent than the last.

Free. No form. No email.

Companion Resource

How the work moves while you build.

This playbook covers what to build and in what order. Its companion covers how that work travels between your AI projects — without you becoming the courier between your own sessions.

Pair with this
The AI Work Routing Standard
Where AI work should happen and how it moves between projects without losing detail. One hub, two artifacts per handoff, one sentence to pick it up.
Read the standard →