Why we publish a baseline at all
Seenu Tech sells AI visibility work, so the most honest thing we can do is run that work on ourselves first and show the results in public. This Week 0 log is the starting line. Rather than open with claims about what GEO can do, we are recording where our own business actually stands in AI answers before we change anything. If we cannot make Seenu Tech understandable and citable to ChatGPT, Perplexity, and Gemini, we have no business asking a client to trust us with theirs. Most agencies never publish their own starting point, because a weak baseline is uncomfortable to show. We think that discomfort is exactly the point: publishing it forces us to improve against a number we cannot quietly revise later. It also changes the kind of company we have to be. Every future post in this log has to measure progress against this exact moment, not against a flattering memory of it, and anyone — a prospect, a competitor, a skeptic — can hold us to it. A baseline you are willing to show in public is a very different promise from one you keep in a private dashboard.
What a baseline actually means here
A baseline is not a launch announcement. It is the measured condition of the business at one moment, written down so that improvement can be judged honestly later. For AI visibility that means two different things, and we keep them separate on purpose. The first is the technical state: is the site crawlable, is the sitemap live, is structured data in place so machines can parse the basics, do the pages load and resolve cleanly. The second is the answer state: when a model is actually asked about us or our category, what does it say, and are we present at all. These two do not move together. A site can pass every technical check and still be completely absent from the answer a buyer reads, because the model had nothing specific and trustworthy to pull from. Treating 'the site is live' as 'the site is working' is the most common mistake in this field, and Week 0 is built to avoid it. We record the technical state and the answer state as two separate facts, so that later we can tell which kind of work actually moved the needle.
What the baseline showed
Using only what we could verify, the technical foundations were in place: the public site was live and crawlable, the sitemap and robots file were reachable, structured data was present on the key pages, and the public Living Lab page existed as a fixed reference point a skeptic could open. We also checked the smaller things that quietly break visibility — that important pages were not blocked, that canonical signals were consistent, that the bilingual English and Korean structure resolved correctly. The answer state was a different and more humbling story. Public AI visibility for a young brand is early, and we treated it that way rather than dressing it up. When the major engines were asked broad category questions, we were not a confident part of the answer yet. That gap is not something to hide; it is the entire reason this log exists. A clear, modest starting point is far more useful than an inflated one, because every later change has something concrete and unflattering to be measured against.
The weekly measurement set we defined
We defined five prompt types to recheck every week, in both English and Korean, because each one fails in a different way. Brand prompts — 'What is Seenu Tech?' — test whether our own pages define us clearly enough to be repeated back accurately. Category prompts — 'What is generative engine optimization?' — test whether we appear in the topic at all. Buyer-intent prompts — 'Who can help my New Jersey business show up in AI search?' — are the ones that create real demand, and they are the hardest to win. Vertical prompts check the specific industries we serve, such as commercial real estate or dental clinics, where a single answer can be worth a real client. Expansion prompts watch adjacent topics we want to grow into next. Running the same set on a fixed cadence, in two languages, is what turns a one-time snapshot into a measurement habit, and a habit into a record that a buyer can trust. It also tells us where to spend the next week's effort, instead of guessing.
What we are not claiming yet
We are not claiming any guaranteed placement in AI answers, and we never will, because no honest provider can promise that — the engines decide, not us. We are not reporting citation frequency, rankings, lead volume, or revenue at Week 0, because a single reading cannot separate signal from noise; AI answers vary run to run, so one good response proves almost nothing. Anyone who shows you a dramatic AI-visibility result after a few days is selling a story, not a method. The point of a baseline is restraint: state what is verifiable today, mark clearly what is not, and let the weekly record earn the bigger claims over time. Saying plainly that our public AI visibility is still early is not a weakness in the proof — it is the proof working as intended. When we do report a gain in a future week, this disciplined starting point is what will make that number believable.
What comes next
From here the work is repetitive on purpose: rerun the five-prompt set each week, deepen the pages that models reward with clearer answers and stronger proof, fix the technical gaps we find, and publish what moves — including the weeks where nothing does. Honest logs include the flat weeks, because hiding them would quietly turn a measurement system back into marketing. The public Living Lab tracks the same progress in one place, and the AI Visibility Audit applies this exact baseline process to your business instead of ours, so you get your own starting number to improve against. If you want to follow the operating record as it builds, the Business Blog is where each week lands. Week 0 is intentionally unglamorous; its only job is to be the honest line that everything after it is measured against — and to prove, by example, that we run the same discipline we sell.