Contents
Two worlds called "AEO"
Search "AEO" in 2026 and you'll find 20+ tools, dozens of agencies, and hundreds of blog posts — all focused on one thing: getting your brand mentioned when humans ask ChatGPT a question.
That's Content AEO. It's real, it's growing, and Profound alone raised enough to hit a $1B valuation doing it.
But there's a second AEO that nobody's talking about. When an AI agent receives a task like "create an invoice in freee" or "update the Salesforce pipeline," it doesn't ask ChatGPT for recommendations. It runs a semantic search over installed MCP tools and picks the best match. If your SaaS doesn't show up in that search, it's not just un-recommended — it's un-considered.
That's Product AEO — what we call Agent AEO. And the gap between these two worlds is where most SaaS companies are blind.
Content AEO: Getting mentioned in ChatGPT
Content AEO is the discipline of optimizing web content — blog posts, documentation, FAQ pages — so that AI-powered answer engines cite your brand. The key question: "When someone asks ChatGPT about our category, does our name come up?"
How it works
- AI chatbots (ChatGPT, Perplexity, Gemini, Copilot) generate answers from their training data and search results
- Your visibility depends on content presence in training corpora, search index ranking, and structured data markup
- Optimization targets: blog content, FAQ schema, knowledge base articles, PR mentions
What it measures
- Brand mention frequency in AI-generated responses
- Sentiment of mentions (positive, neutral, negative)
- Share of voice vs. competitors in AI responses
- Citation accuracy (does the AI get your product details right?)
Who does it
Profound ($1B valuation), Peec AI, Semrush (AEO module), Ahrefs (AI Overview tracking), Conductor, BrightEdge, Otterly, and a growing roster of agencies.
Product AEO: Getting selected by agents
Product AEO (Agent AEO) is the discipline of optimizing your SaaS product itself — its MCP server, tool schemas, API descriptions — so that AI agents discover and select it when performing tasks. The key question: "When an agent needs a tool that does what we do, does it find us?"
How it works
- AI agents (Claude Code, Cursor, custom agent workflows) search installed MCP servers using semantic tool search
- Your visibility depends on MCP server name, tool descriptions, parameter schemas, and success track record
- Optimization targets: MCP server metadata, tool
descriptionfields,inputSchemadefinitions, llms.txt / mcp.json
What it measures
- Discovery rate: Does Tool Search return your server for relevant queries?
- Selection rate: When discovered, is your tool chosen over alternatives?
- Success rate: Does the agent complete the task after selecting your tool?
- search_miss rate: How often does your service fail to appear at all?
Who does it
KanseiLink. That's the entire list as of mid-2026.
The complete comparison table
| Dimension | Content AEO | Product AEO (Agent AEO) |
|---|---|---|
| Who searches | Humans asking AI chatbots | AI agents performing tasks |
| Search mechanism | LLM knowledge retrieval + web search | Semantic tool search over MCP schemas |
| What you optimize | Web content (blogs, docs, FAQs) | Product interfaces (MCP server, tool schemas, APIs) |
| Success metric | Brand mentioned in AI response | Task completed using your service |
| Failure mode | Competitor mentioned instead | search_miss — not even considered |
| Failure visibility | Visible (you can see competitor mentions) | Invisible (no log, no alert, no trace) |
| Update cadence | Slow (LLM training cycles, months) | Instant (schema changes take effect immediately) |
| Market maturity | 20+ tools, $200M+ invested | 1 tool (KanseiLink), nascent |
| Key tools | Profound, Peec AI, Semrush, Ahrefs | KanseiLink |
| Analogous to | SEO for Google (content ranking) | ASO for App Store (product listing) |
Why the mechanisms are fundamentally different
It's tempting to think of Product AEO as "the same thing but for APIs." It's not. The mechanisms are different at every layer.
Different search engines
Content AEO targets LLM knowledge — what the model "knows" from training data plus real-time web search. Product AEO targets Tool Search — a separate semantic search system that operates over locally installed MCP server metadata. These two systems share no data. A product with great web presence but a poorly-named MCP server will be mentioned in ChatGPT answers but invisible to agents doing actual work.
Different optimization surfaces
Content AEO optimizes text on the open web: blog posts, FAQ pages, documentation. Product AEO optimizes machine-readable interfaces: the name and description fields of your MCP server, the description and inputSchema of each tool, and emerging standards like llms.txt and mcp.json. One is for humans reading; the other is for machines parsing.
Different feedback loops
Content AEO failures are visible — you can monitor AI chatbot responses and see when competitors get mentioned instead. Product AEO failures are invisible. When an agent's Tool Search doesn't return your MCP server, there's no log entry, no error, no HTTP request to your server. The failure exists only in the agent's decision space, where KanseiLink tracks it as search_miss.
Why your SaaS needs both
Research Phase
- Human asks: "Best CRM for startups?"
- ChatGPT mentions your brand
- Content AEO drives this
- Leads to awareness + consideration
Execution Phase
- Agent receives: "Update pipeline in CRM"
- Tool Search finds your MCP server
- Product AEO drives this
- Leads to usage + retention
These are two different decision points in the customer journey. Content AEO captures the research moment — when a human is evaluating options. Product AEO captures the execution moment — when an agent is performing a task.
A SaaS company that invests only in Content AEO will be recommended by chatbots but not selected by agents. A company that invests only in Product AEO will be selected by agents but not discovered by humans researching options. The complete AI visibility strategy needs both.
Think of it like SEO vs. ASO. SEO gets you found on Google. ASO gets you found in the App Store. Both drive growth, but through different mechanisms. Content AEO is the "SEO" of AI visibility. Product AEO is the "ASO." You need both, and they require different teams, different tools, and different optimization strategies.
The tool landscape: Who covers what
| Layer | What it tracks | Key players |
|---|---|---|
| Content AEO | "Is our brand mentioned in ChatGPT?" | Profound ($1B), Peec AI, Semrush, Ahrefs, Conductor, BrightEdge, Otterly |
| MCP Directories | "Is our MCP server listed?" | Official Registry, Smithery, Glama, PulseMCP, mcp.so |
| MCP Internal Monitoring | "Is our MCP server healthy?" | MCPcat (session replay), Speakeasy (gateway), Tinybird |
| MCP Benchmarks | "What's our score on standard tests?" | mcpbr.org, Microsoft MCP Interviewer |
| Product AEO (Agent AEO) | "Do agents find and select our product?" | KanseiLink |
Every tool in the Content AEO row answers a valid question. But none of them answer the question that matters for the agent economy: "When an agent needs what we sell, does it find us?"
What to do now
If you've invested in Content AEO but not Product AEO
- Check your agent visibility: Search your service on KanseiLink to see your AEO grade and how agents discover your product
- Publish an MCP server if you haven't. API-only services grade markedly lower on agent readiness than official MCP services in KanseiLink's evaluation (success-rate measurement data is being accumulated)
- Optimize tool descriptions for intent-based search. Agents search "create an invoice" not "Acme MCP Server"
- Deploy llms.txt at your domain root — a simple text file that helps AI models understand your service
If you've invested in neither
- Start with Content AEO — it's faster to implement (publish structured content, add FAQ schema) and the tools are mature
- Plan for Product AEO in parallel — the agent economy is growing at 73,000+ MCP servers and 97M SDK downloads/month
- The first-mover advantage in Product AEO is massive. Most categories have zero or one MCP-optimized provider
FAQ
What is the difference between Content AEO and Product AEO?
Content AEO optimizes web content so AI chatbots mention your brand. Product AEO optimizes your SaaS product itself — MCP servers, tool descriptions, API schemas — so AI agents select your service when performing tasks. Different mechanisms, different metrics, different tools.
Why do SaaS companies need both?
Content AEO captures the research phase (human asks "which CRM is best?"). Product AEO captures the execution phase (agent receives "update the pipeline"). These are different decision points. A SaaS visible in ChatGPT but invisible to agent tool search loses the execution layer entirely.
Which AEO tools track agent tool selection?
As of mid-2026, only KanseiLink. All other AEO tools (Profound, Semrush, Ahrefs, etc.) track content-side brand mentions in chatbot responses. The Product AEO space is where Content AEO was in 2023 — the problem exists, but the tooling hasn't caught up.
Can Content AEO improve my agent discoverability?
Not directly. Agent tool selection uses semantic search over MCP schemas, not LLM knowledge retrieval. A product with excellent Content AEO but no MCP server is still invisible to agents. However, strong brand presence helps when users explicitly name your product in agent instructions ("use freee to create invoices").
Is Product AEO only for companies with MCP servers?
No. Product AEO also covers llms.txt (live standard), mcp.json (draft spec), Google ARD (early stage), and OpenAPI specification optimization. But having an MCP server is the highest-impact step — it's the primary way agents discover and use SaaS tools.