Let me tell you what happened on my screen earlier today.
I opened Claude. I typed a prompt. I said: “Connect to my SEOABLE account, pull the analytics and search data for seoable.dev, find the real opportunities, build me a content strategy, draft three blog ideas, and then make me a strategy dashboard and a Word document I can share with my team.”
Then I hit return.
Claude did not write me a generic strategy. Claude did not hallucinate metrics. Claude did not give me the usual confident-but-vague essay about “leveraging your SEO funnel.”
Claude called SEOABLE.
It opened a connection to my Google Analytics account. It pulled the real session numbers for the last ninety days. It pulled my Search Console data. It identified that my site, the company I built specifically to teach founders about AI search visibility, received exactly eight organic Google sessions in the last thirty days. Eight. Not eight hundred. Eight.
Then it pulled my top queries. Found that I am ranked at position 72 for “scalable SEO platform” with 75 impressions. Found that I am at position 18 for “SEO for Bubble apps” with almost no effort. Looked at where my traffic actually comes from. Saw that paid Facebook is doing all the work and direct traffic — podcast listeners, Substack readers — is the only segment that engages meaningfully.
Then Claude built me a three-pillar strategy. Drafted three content ideas, each one targeting a query I’m already indexed for. Generated a dashboard visualization with the real numbers. And produced a complete strategy document in Word format that I can hand to a designer, a writer, or a co-founder right now.
Total time, start to finish, was about ten minutes.
I never logged into SEOABLE. I never opened Google Analytics. I never copied a URL into a dashboard. I never switched tabs.
I asked. The assistant did the rest.
Episode 3 of Cited is about what just happened, why it matters, and why every B2B SaaS company on earth is about to be reorganized around this exact pattern.
The shift
For thirty years, every B2B SaaS company on earth has been organized around the same idea. Build a dashboard. Train users to log in. Optimize the interface. Charge a subscription.
That model worked because the operator of the software was a human. The human opened a browser tab. The human navigated to your URL. The human learned your interface. The human pressed your buttons.
That operator is changing.
A whole generation of founders, especially technical founders, are no longer using software the way the previous generation did. They are not opening fifteen browser tabs. They are not learning fifteen different dashboards. They have a primary assistant. Claude. ChatGPT. Cursor. And the work happens in a conversation with that assistant, not in a series of interfaces.
For these founders, “log into our dashboard” is friction. For agentic workflows, where the assistant itself is trying to do the work, “log into our dashboard” is a non-starter. You cannot ask an autonomous agent to navigate a UI, click around, copy data, and paste it back into a chat. That is not how any of this works.
What the assistant needs is a way to call your product directly. Like a function. Like a tool. Like a capability it can summon when the work calls for it.
There is a name for this now. MCP. Model Context Protocol.
If you have ever asked Claude to search the web, that is a tool call. Claude does not have the internet inside it. It calls an external tool, gets a result, and uses it in the answer. MCP just opens that pattern up. Any company can ship an MCP server. Any MCP server lets an assistant call it the same way it calls web search. Inside the conversation. As a native capability.
That is the protocol. That is the standard. That is the door that just opened.
The companies that ship MCP servers in 2026 are the companies that will live inside the next generation of work. The companies that do not are about to find their dashboards getting very quiet.
This is not a “nice to have.” This is the new “have a mobile app.” Companies that did not take mobile seriously between 2008 and 2012 are mostly dead now. Same logic. Different shift. Bigger stakes.
Why I built MCP into SEOABLE
When the assistant integration requests started coming in from customers, I had to make a decision. The easy path was to ship an API. APIs are well understood. Every B2B SaaS company has one. APIs let developers integrate.
But APIs put the burden on the developer. The developer has to write code. The developer has to handle authentication. The developer has to glue the API into their workflow. APIs are how you integrate with a system you control.
MCP is different. MCP lets the assistant integrate without the developer. The user just asks. The assistant figures out the rest. The friction goes from “write some code” down to “type a sentence.”
So I shipped both. SEOABLE has an API for the developers who want it. SEOABLE also has an MCP server, which is the part I am most excited about. Because the MCP server is what lets any founder using Claude or Cursor or any MCP-compatible assistant ask their assistant to do real SEO and AEO work, with real data, in real time, without ever opening a dashboard.
What happened earlier today, the workflow I just described, is the proof.
What just happened, mechanically
I want to walk through it one more time, slowly, because the mechanics matter.
I am sitting in Claude. Not Claude Code, not Cursor, just Claude on the web. I want a content strategy for SEOABLE based on what is actually working and what is not. I do not want to log into Google Analytics. I do not want to log into Search Console. I do not want to log into SEOABLE either.
I type a prompt. I hit return.
Claude does not write me a paragraph of advice. Claude pauses. Then a little indicator appears on screen. It says “using SEOABLE tools.” Claude is making a tool call.
Behind the scenes, the SEOABLE MCP server has just been pinged. SEOABLE is asking my connected Google account for the analytics data. The data comes back. Real session counts. Real engagement rates. Real traffic sources. Real top pages. The numbers are not made up. They are not estimates. They are the actual numbers from my actual account, pulled live, inside the conversation.
Claude reads the data. Then it calls Search Console. Same flow. SEOABLE asks Google. Google returns the real queries my site has been showing up for. The real impression counts. The real positions. Claude reads that too.
About forty seconds later, Claude starts to write.
Here is where it gets interesting. Because Claude does not write what an AI would write if it was guessing. Claude writes what an analyst would write if they had been studying my account for an hour.
The first thing it tells me is the thing I most needed to hear. My site received eight organic Google sessions in the last thirty days. The SEO and AEO company has almost no SEO of its own.
That is not a hallucination. That is the number, pulled from the API, in front of me on screen.
Then it tells me something even more useful. I am ranked position 72 for “scalable SEO platform” with 75 impressions over 90 days. Page 7 of Google. But indexed. The keyword exists. The opportunity is real. I just have no page actually targeting it.
Then it tells me about a query I had completely forgotten about. “SEO for Bubble apps.” Position 18. Two impressions. Zero effort. That is a signal. That is Google telling me, in its own quiet way, that there is an underserved niche there waiting for me to take it.
Then Claude builds the strategy.
Three pillars. Pillar one: win the queries I am already showing up for. Ninety-day payoff. Pillar two: own the no-code SEO niche. Six-month payoff. Pillar three: make this podcast the engine. Twelve-month payoff.
For each pillar, concrete action items. Build this landing page. Add this schema. Replicate this pattern.
Then Claude drafts three content ideas. Each one tied to a real query in Search Console. Each one with target keywords, an angle, a word count, and a recommended deadline.
Then it generates a dashboard. Live, on screen, rendered inside the chat. Real numbers. My numbers. Visualized. With buttons that let me drill in further.
Then it produces a Word document. A polished, properly formatted, eight-page strategy doc I could hand to a writer or a co-founder right now and have them start working from.
All of that happened in about ten minutes.
Why this matters more than it looks
There is a temptation to read what I just described and treat it as cute. A neat demo. Look what AI can do.
It is not cute.
What just happened is the entire dashboard, the entire reporting layer, the entire “go log in and run a report and copy the numbers and paste them somewhere and write a doc and share it with your team” workflow that has dominated SaaS for three decades, collapsed into a single sentence.
I did not open Google Analytics. I did not open Search Console. I did not open SEOABLE. I did not write a single line of code. I did not click a single button. I asked. The assistant called the right tools. The tools returned the right data. The assistant turned the data into a strategy. And it handed me a deliverable.
What just happened, strategically, is the entire dashboard collapsed into a sentence.
That is the deal. That is the new contract between software and the people who use it.
For thirty years, software was a destination. You went to it. You learned it. You operated it.
Now software is a capability. The assistant calls it. The assistant operates it. You just ask.
If your product is not a tool an assistant can call, your product is becoming a tab that nobody opens.
That is not a future scenario. That is a Tuesday afternoon, in May 2026, in real workflows, with real founders, right now.
What this means if you are building anything
The shift I have just described is going to define the next five years of software.
If you are building a SaaS product right now, in any category, ask yourself one question. When a founder using Claude or Cursor or ChatGPT wants to do the job your product solves, can the assistant do it without leaving the conversation? Can the assistant call your product directly? Can the assistant pull your data, use it, act on it, integrate it into whatever else the user is doing?
If the answer is no, you are in the same position as a software company in 2009 that did not have a mobile app. You can keep going. You can survive. But every day, a little more of your audience is shifting to a surface you do not exist on. And the longer you wait, the more invisible you become.
The same question applies to every category. Project management. CRM. Analytics. Email. Design. Customer support. Billing. Knowledge bases. Note-taking. Calendar. Every single one. They are all about to be reorganized around the assistant.
Some will get there fast. Some will get there slowly. Some will get there too late.
The companies that get there fast become infrastructure. They become the tools the assistant trusts. They get called. They get cited.
The companies that get there too late become the websites the assistant used to recommend, before users stopped clicking through.
Why SEO and AEO have to live inside the assistant
Most SEO work is not interface work. It is workflow work. You audit a page. You generate a content plan. You write a blog post. You add schema. You check if the changes worked.
None of that requires a dashboard. All of it requires a capability that can be called when it is needed.
And almost all of it is something founders are already trying to do inside their assistant. They are already asking Claude to write blog posts. They are already asking ChatGPT what schema to add. They are already pasting URLs and asking for audits.
They just do not have a real tool behind the answer. They have whatever the assistant guesses.
When you ask Claude or ChatGPT to audit your SEO without a tool behind it, you get a confident, plausible answer. Sometimes it is right. Often it is slightly wrong. Occasionally it is completely fabricated. Because the model is reasoning, not checking. It is generating an audit, not running one.
That is the gap SEOABLE fills. The MCP server gives the assistant a real tool to call, with real data, real schema parsing, real citation analysis, real audit logic.
The assistant stops guessing. It starts knowing.
That is the difference.
The founders who connect SEOABLE to their assistant in the next few weeks are early to a curve every founder in their space is going to be on by next year. Maybe sooner.
Three things to do this week
One. Open Claude, ChatGPT, or whichever assistant you use most. Ask it to do something you would normally do inside a dashboard. Anything. Pull your last month of email opens. Check your Stripe revenue. Audit your competitor’s homepage. See what happens. See how close it gets without a real tool. Then ask yourself which tools you wish were connected. That list is your map.
Two. If you are a founder running a SaaS product, audit your own roadmap. Is there an MCP server on it? If yes, when? If no, why not? The companies shipping MCP in the next 90 days are going to be the companies that look like infrastructure by the end of the year. The companies that are not are going to look like tabs.
Three. If you are a SEOABLE customer, or thinking about becoming one, the MCP server is in early access right now. Go to seoable.dev. Sign up for the early access list. Connect it to your Claude account. Run an audit by typing a sentence. See what it feels like when the assistant has a real tool behind the answer. Once you have used it that way, you will not want to go back.
What’s coming next
Episode 4 of Cited drops next week. The schema episode. Why structured data is suddenly the most important AEO lever, and the exact JSON-LD blocks every founder should add to their site this weekend. Tactical. Specific. The most useful 30 minutes you will spend on AEO this year.
But Episode 3, this one, is the one I would send to one founder right now if I could. Because if you are building anything, in any category, the shift I just described is going to define the next five years of software.
The companies that understand it will compound. The companies that do not will quietly disappear.
The game is inside the assistant.
Andrew
Founder, SEOABLE
Host, Cited
Cited is the SEOABLE podcast. Weekly, founder-to-founder, about getting found in the AI era.
