A business owner in Kuala Lumpur asked us a fair question earlier this year: he had two proposals on his desk, one from an SEO agency quoting a monthly retainer, one selling an "AI SEO" service for a fraction of it, and he wanted to know whether the cheap one was a worse version of the expensive one or a different thing entirely.
It is a different thing, but not in the way either proposal explained. And a good deal of what gets sold as "AI SEO" in 2026 is a rebrand rather than a difference — so it is worth being precise about what genuinely changes.
What traditional SEO actually sells you
Strip away the language and a traditional SEO retainer sells a specialist's attention. You pay for a number of hours a month, and someone competent decides where those hours go — an audit, some content, a technical fix, a bit of link outreach.
That model has a structural consequence people rarely name: the work happens in bursts, because attention is expensive. Your site gets audited in month one, largely ignored in month two while the specialist works someone else's account, and revisited in month three. The retainer buys capacity, and capacity gets rationed.
It is not a scam. For a lot of businesses it is the correct purchase, and I will come to exactly when. But it explains why SEO retainers feel lumpy, and why the first month always looks the most productive.
What actually changes with AI SEO
Two different things change, and almost every vendor conflates them. Separating them is most of what you need to evaluate a proposal.
The first change is how the work gets produced. An agent monitors the site continuously rather than in scheduled bursts. It crawls, watches Search Console and analytics, finds the technical faults and the content gaps, and produces a specific proposed change — this page, this title, this reason, this expected effect. A human approves, rejects, or edits it. Nothing goes live unapproved.
The consequence is not that the work is smarter. It is that small, frequent, unglamorous work becomes affordable. Fixing a broken canonical, tightening a title that undersells a page, adding an internal link from a page that ranks to a page that does not — none of that is hard, all of it matters, and all of it is uneconomic to do at human rates every week.
The second change is the one people get wrong, and it is worth a section of its own.
The answer-engine question, and an honest answer
The pitch you will hear is that search has changed — AI Overviews, ChatGPT, Perplexity — and that you now need "AEO" or "GEO" services, special markup, and an llms.txt file to be visible in AI answers.
Here is what Google's own documentation says about appearing in AI Overviews and AI Mode:
"There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary."
And, more pointedly:
"You don't need to create new machine readable files, AI text files, or markup to appear in these features."
And on schema:
"There's also no special schema.org structured data that you need to add."
What Google says you actually need is ordinary eligibility: the page must be indexed and able to show with a snippet, meeting the normal Search technical requirements.
So if someone is selling you a Google-AI-visibility package built on special files and markup, they are selling something the platform explicitly says is unnecessary. That is worth knowing before you sign.
Our own honest position, since we do maintain llms.txt files. We treat them as a low-cost hedge aimed at the non-Google answer engines — ChatGPT, Perplexity, Claude — which have published no equivalent statement in either direction. Nobody can honestly tell you it works for them, because nobody has said. It costs very little to keep a clean, accurate summary of your site, so we do it and we do not charge a premium for it. What we will not do is tell you it gets you into Google's AI Overviews, because Google has said it does not.
The real answer to the answer-engine question is duller and more useful: being cited by an AI system is downstream of being genuinely good on the ordinary things. Clear pages that answer real questions, correct technical foundations, and a site that a crawler can understand. The work did not change as much as the marketing did.
Where traditional SEO still wins, and it genuinely does
I would rather tell you this than have you find out.
Link acquisition and digital PR. Earning a link from a publication or an industry body is relationship work — pitching, negotiating, knowing who to call. That is human, and an agency with real media contacts in Malaysia is buying you something an agent cannot produce.
Complex technical migrations. Replatforming, large-scale redirect mapping, international or multi-language site architecture. These need a specialist who can hold the whole system in their head and take responsibility when something breaks at 11pm. Do not run a migration through an approval queue.
Commercial strategy. Deciding which market to enter, which service line deserves investment, whether a keyword's traffic is worth having at all — that is business judgement informed by search data, not derived from it.
If your bottleneck is one of those three, hire a good human and pay properly for them.
Where the agent model wins
Consistency. Four small proposals a month, every month, applied. Most sites lose ground to neglect rather than to competition, and neglect is exactly what continuous monitoring prevents.
Cost, and it is not marginal. Our SEO Agent Essential is RM699 a month — one website, up to 10 tracked keyword themes, roughly four technical proposals a month with one-click approval, weekly technical and indexing checks, and a monthly report. SEO Agent Growth is RM1,499, adding up to 25 keyword themes and a blog post every week. Both are billed month to month with no lock-in.
For context on the Growth tier: our own estimate is that a freelance SEO writer in Malaysia runs roughly RM300–800 per post, so four posts a month is somewhere between RM1,200 and RM3,200 of human-equivalent output. That is an estimate from our own costing rather than market research, but it is the arithmetic that makes the tier make sense.
Volume of small fixes. The unglamorous backlog that never gets prioritised in a human retainer because it is not worth a specialist's hour.
One genuine constraint, stated plainly: applying changes needs source or CMS access. Insight works on any website; making the change only works where we have the keys. Sites we built are cheapest to onboard, which is what the external-site onboarding fee reflects.
What neither can promise
Neither approach can guarantee a ranking. Nobody outside Google controls Google's ranking systems, and any provider guaranteeing you position one is either misunderstanding the mechanism or hoping you will not check.
What a competent programme can promise is the work: faults found and fixed, pages that answer real queries, internal links that make sense, and a written record of what changed and why. If a proposal in front of you promises an outcome instead of a process, that is the thing to question.
So which should you buy
Roughly:
- Your site is technically messy, nobody has touched it in a year, and you need steady maintenance — the agent model, and the cost difference is large enough to be decisive.
- You need links, a migration, or a strategic call — a human specialist, properly paid.
- You need both — that is common, and they are not mutually exclusive. Run the agent for the continuous work and buy human time for the campaigns.
The question to ask any provider, in either camp, is the same one: show me exactly what you will do in month two. The answer separates the serious from the rest faster than any comparison of methods.
Konsultasi percuma
Not sure which one your site actually needs
Send us the URL. We will tell you whether the bottleneck is maintenance, links, or something structural — including when the honest answer is that you need a human specialist rather than us.