An AI SEO system, and the eleven jobs inside it

What an AI SEO system actually is

An AI SEO system is a group of AI workers, each responsible for one part of getting found, all sharing one set of facts about your business. Getting found is not one job. It is research, planning, writing, fixing pages, keeping your listing on the map correct, explaining your business to machines in a format they can read, then watching it all weekly. A tool that claims to do SEO usually does one of those seven things and leaves the other six to you.

Search itself has split into several places at once. Somebody looking for a supplier might type a phrase into Google, might read whatever answer box appears above the results without clicking anything, or might ask an AI chat tool which company to use and take its word for it. Those are different jobs with different rules, which is why the roster below has separate workers.

None of it is a trick and none of it is fast. What AI changes is the continuity: the research does not stop because somebody went on leave, and the monitoring does not stop because it is boring.

The team, and what each one does

Read this as a staff list. If you cannot say in one sentence what a worker does, it should not be in your system.

• The keyword researcher. Collects the phrases people actually type when looking for what you sell, and separates the ones typed by buyers from the ones typed by students and job seekers.
• The content planner. Turns that list into a plan: which pages exist, which need rewriting, which are missing, and in what order so the useful ones come first.
• The writer. Produces the drafts. Buying guides, comparison pages, answers to the questions your sales team is tired of repeating, each built from a brief rather than from nothing.
• The page optimiser. Works on pages you already have. Clearer headings, an answer near the top instead of four paragraphs of throat clearing, and a plain statement of who you serve.
• The question answerer. Finds the questions people ask out loud or type as full sentences, and makes sure a short, direct answer to each sits on a page.
• The AI search explainer. Deals with AI assistants recommending suppliers, which mostly means stating your facts plainly in one place rather than scattering them across a brochure.
• The schema writer. Writes the hidden labels that tell a machine which text is your address, which is a price and which is a review. Google stopped showing FAQ rich results on 7 May 2026, so this is not about decoration in the listing. It is about being read without guesswork.
• The local listings keeper. Looks after your Google Business Profile. Opening hours that are right, photographs that are not five years old, and a nudge when a review needs a reply.
• The location page builder. If you serve several areas, it builds a real page for each, with the differences that make it real rather than the same page with the town name swapped.
• The rankings monitor. Checks positions on a schedule and separates a genuine fall from the daily wobble that means nothing.
• The competitor watcher. Tracks what your rivals publish and reports the gaps: the questions they answer that you do not.

A Tuesday inside an AI SEO team

6:30am. The rankings monitor runs before anyone is awake. Most days it finds nothing worth saying. Today one page has dropped several positions for a phrase that brings real enquiries, so it checks whether the whole site moved or just that page. Just that page. It hands the finding on.

8:00am. The page optimiser reads the page that dropped and compares it to the three now above it. Two of them answer a question yours does not answer at all. It drafts an extra section and puts it in the approvals lane, because publishing on your website is a decision a person signs off.

9:45am. The keyword researcher finishes its weekly sweep and finds a phrase that has started appearing: people searching for your service with the word urgent in front of it. That is not really a keyword, it is a fact about your customers, so it goes to the content planner and to the sales team.

11:15am. The local listings keeper spots a review that came in yesterday afternoon with nobody replying. It drafts a reply, does not post it, and puts it in the lane for approval. Replies to real customers with names attached are not something a machine sends unsupervised.

12:30pm. The question answerer has been reading what people ask on your listing and in your inbox. Three are the same question about delivery to one area, and no page answers it. It writes one and adds it to the plan.

2:00pm. The schema writer checks the hidden labels on last week's pages: is the address on the contact page marked as an address, do the opening hours in the code match the ones on your listing. It finds a mismatch, the sort of thing a person would never notice and a machine reading your site would.

3:30pm. You spend eleven minutes on approvals. Two drafts, one review reply, one page change. You reject one because the tone is not yours, and the correction is kept.

4:45pm. The competitor watcher reports that a rival has published a comparison page naming your category. It recommends nothing, because that is a judgement about your business. It puts the fact in Friday's summary, alongside what moved, what was published and one line on what to do next.

What stays human in search work

Publishing stays human. Anything on your website carries your name and can be quoted back at you by a customer or a regulator, so a person reads it first. Approving a page takes about as long as reading it.

Claims stay human. A price, a guarantee, a delivery time, a professional credential. An AI worker told these facts repeats them correctly, and one that has not been told fills the gap with something plausible. So the facts live in one place a person maintains, and anything not there is not stated.

Replies to named customers stay human. A review from a person with a name, especially an unhappy one, is a relationship rather than a content task. Drafting is fine. Sending without reading is not.

Judgement about what to chase stays human. Some phrases bring traffic and no customers, and you already know which enquiries waste your week. That is in no keyword tool.

How search work connects to other departments

Search sits between three other departments, and those handoffs are the reason it belongs in an operating system rather than at an agency somewhere else.

Content is the obvious one. The planner decides what to write and why, the content team writes it, and the loop closes when the rankings monitor reports whether the piece did anything. An agency arrangement usually breaks that loop, because the people writing never find out what happened. The other two are cheaper and almost nobody wires them up.

• The content team writes what the planner briefs, and hears back whether it worked.
• The marketing team hands over the phrases that already convert in paid, which is the best keyword evidence you can own because you paid to find it out.
• The support team passes on its most repeated questions, which are word for word what people type into search.

What an AI SEO system costs and what changes the number

It is an audit, then a build, then a monthly fee while it runs. There is no figure here because the same roster is very different work in two businesses that look similar from outside. A five page website in one city with a clean Google listing is not the same job as a hundred page site across four service areas.

The largest variable is how much of your website you are allowed to change. Where somebody at your end can edit pages directly, work happens weekly. Where every change is queued with a developer who has other clients, the system waits and you pay for a plan sitting in a queue. The second variable is how many of the eleven workers you need.

• How many pages exist now, and how many of them are worth keeping.
• Whether someone at your end can publish changes without a queue.
• How many locations or service areas need their own real page.
• Whether writing is included or your own team drafts and the system edits.
• How competitive the phrases are, which decides how long before anything moves.
• Whether Wobble operates it monthly or hands it to your team once the handover is done.

When search is not the thing holding you back

If nobody answers the enquiries you already get, more visibility makes the leak bigger. Businesses here describe losing orders because a message sat unanswered for hours while the customer went elsewhere. Sending more people into that is paying to lose more sales. Fix the answering first.

If you need customers this month, this is the wrong lever. Pages take time to be found, trusted and ranked, and no amount of AI shortens that. Paid advertising buys attention today and search earns it later, so an urgent gap is filled by the first while you build the second.

If your website cannot take an enquiry cleanly, ranking better sends more people to a page that does not convert them. Check how many reached your contact page last month and how many actually got in touch.

If buying in your market happens on referral, search may never be your channel. Some trades genuinely work that way, and being told so is more useful than being sold a plan. Ask your last ten customers how they found you, and if none of them typed anything, believe them.

Who approves, what moves the cost, and who owns the pages afterwards

The approvals lane in that Tuesday is not decoration. Nothing is published in your name without human approval, and that covers the page draft, the review reply, the schema change and anything a customer with a name attached will read. Where a worker finds something it cannot judge, it hands it to a person with the finding attached rather than recommending an action, which is why the competitor watcher reports a fact and then stops. A system publishing unsupervised will eventually publish something about your business that is not true, and the cost of that lands on you rather than on whoever built it.

There is no price on this page. What moves the number is how many pages already exist and how many of them are worth keeping, whether you serve one location or eleven, how much research can run against your own search console data rather than being bought in, and who does the approvals every week. The last of those is a cost you carry rather than pay, and it is roughly the eleven minutes described above. A supplier quoting on article count has priced the writing and ignored the other ten jobs.

The output is yours in a way that matters more here than elsewhere, because search work is cumulative and a supplier holding it holds your visibility with it. The pages sit on your own domain. The Google Business Profile stays in your account. The keyword research, the briefs and the rankings history live in accounts under your logins, and the workers themselves are workflow definitions rather than a subscription to somebody's product. If you leave, you own the system and every page it produced.

Moiz Khan owns automation architecture at Wobble and decides what a worker may do without asking. Ali owns marketing and sales, which is the department this system exists to serve, and Haad owns growth and client solutions. Wobble works from Karachi, bills month to month and is answerable for what it operates, across 25 engagements in six countries. The published search work nearest to this is Zavcom, a United States internet provider, at 89 conversions in ninety days at $1.54 a click. That is paid search rather than organic, and the distance is worth stating rather than letting it pass.

Common questions

What is an AI SEO system?

It is a team of AI workers, each doing one job in search visibility, connected to your website and to each other. Keyword research, content planning, page optimisation, structured data, local listings and rankings monitoring are separate jobs with separate workers rather than one tool called SEO. A person approves anything published.

Is this different from an SEO tool I could buy myself?

A tool reports. It tells you where you rank, then waits for a human to act. Here the acting is included: planning, drafting, page changes and listing maintenance are jobs with their own triggers, so reporting turns into work rather than a dashboard nobody opens.

How is AEO different from SEO, in plain terms?

SEO is about being one of the results a person clicks. AEO is about being the answer read out or shown in a box, often with no click at all. The practical difference is format: AEO rewards a short, direct answer near the top of a page, phrased as a person would ask it, rather than a long article that gets to the point eventually.

How long before an AI SEO system shows results?

Longer than anyone selling it would like to say. Technical corrections and local listing work can move things within weeks. Ranking for competitive phrases with new pages takes months, because the delay is in how search engines confirm a page is worth trusting rather than in how fast it was written.

Will AI written pages be penalised by Google?

The stated position is that search engines judge whether a page is useful rather than how it was typed. In practice, thin pages produced quickly do badly whether a human or a machine wrote them. The pages that hold up carry real information: your process, your service area, the questions you get asked.

Do I need all eleven of these AI workers?

Almost nobody starts with all of them. A single location business selling one service usually needs the keyword researcher, the page optimiser, the local listings keeper and the rankings monitor. Installing the whole roster on day one produces more output than you can approve, which is how these systems get abandoned.