Everyone's talking about AEO.
What's missing: there's no standard way to tell an AI system exactly how to represent your brand.
No file it can read at the source. No spec it can pull before generating a response that mentions you.
Until now.
The problem with how brands exist in AI search today
When an AI answer engine cites your brand, it's working from whatever it can find - your website, press mentions, schema markup, third-party content. It synthesises all of it and generates a description.
Sometimes that description is accurate. Sometimes it's stale. Sometimes it conflates you with a competitor. Sometimes it uses language you'd never use yourself.
You have no direct input into that process.
That's the gap AEO.md fills.
What is AEO.md (Taste.md?)
AEO.md (Taste.md) is a plain-text markdown file you place at your domain root.
It tells AI systems, content agents, and answer engines:
- Who you are - in exact language you want cited
- What questions your brand should answer
- What products and services you offer
- What you are NOT (disambiguation matters more than most people realise)
- What claims you make and where they're sourced
- How your brand voice works
- What trust signals exist to verify your authority
Think of it as the brand layer that sits alongside your existing technical stack:
robots.txt - tells crawlers what to index
llms.txt - summarises your site for AI models
DESIGN.md - tells dev agents how to build your UI
AEO.md - tells AI how to represent and cite your brand
Same pattern. Different layer.
Why this works
Answer engines are probabilistic. They pull from multiple sources and synthesise. The more consistent and clear your brand signal is across those sources, the more accurately you get represented.
AEO.md doesn't guarantee anything - no single file does. But it gives AI systems a first-party source of truth they can weight against everything else they find.
Schema markup is already doing part of this job. llms.txt is doing another part. AEO.md completes the picture by adding the brand voice, canonical Q&A, and disambiguation layers that neither of those formats covers.
It's also useful for your own content agents. If you're using AI to generate blog posts, social content, or sales copy - pointing your agent at your AEO.md before it writes means every output is grounded in your actual positioning, not a hallucinated version of it.
The spec
AEO.md has eight sections:
1. Brand identity - your canonical one-liner and extended description, in the exact language you want cited
2. Category and classification - what you are, who you serve, and critically who you don't serve
3. Canonical questions - the specific questions your brand should be the answer to, with direct answers under 60 words each
4. Key claims - specific, sourced facts AI should use when citing you
5. Products and services - each offering mapped to an audience and a differentiator
6. Brand voice rules - preferred and banned language for content agents
7. Trust signals - press, awards, certifications, founding credentials
8. Disambiguation - what you are not, to prevent AI conflating you with similar entities
How to implement it
- Download the blank template
- Fill it in - most sections take 15-30 minutes if you know your brand
- Place it at yourdomain.com/AEO.md
- Reference it in your llms.txt file:
AEO: https://yourdomain.com/AEO.md - Keep it updated - answer engines have a recency bias, stale specs lose weight
No plugins. No platform dependency. Just a text file.
This is version 0.1
We're calling this 0.1 because we expect it to evolve. As answer engines develop, as llms.txt matures, as agentic AI becomes more common - the spec will need to grow.
What we're proposing today is a starting point and a naming convention. The format can be debated. The need is real.
If you build with it, extend it, or find gaps in it - we want to hear from you.
Why we built this
We work at the intersection of commerce infrastructure and AI operations. Our clients ask us constantly how to stay visible as search shifts from Google rankings to AI citations.
The honest answer was: the tools are fragmented and the spec doesn't exist yet.
So we wrote it.
Flux is a Shopify Plus Agency for Advanced Engineering & AEO
- AEO advice tells you what to implement - schema, FAQs, structured data. It doesn't give AI a first-party source of truth for how to represent your brand.
- AEO.md is a plain-text markdown file you place at your domain root. It tells AI systems exactly who you are, what questions you answer, what you're not, and how to cite you.
- Same pattern as DESIGN.md (UI spec for dev agents) and llms.txt (site summary for AI crawlers) - but for brand citation.
- The spec is open. Drop it in your project root, reference it from llms.txt, and any AI agent or answer engine that reads it has a first-party source.
