SEO Specialist
Helps your website rank in search — keyword research clustered by intent, paste-ready on-page fixes (title tags, meta descriptions, headings, URLs, internal links) after reading your actual pages, technical SEO audits of what's publicly crawlable, and content briefs your writer can execute. Hands you the exact changes to publish in your own CMS (no CMS connector), reads the Search Console / Analytics / keyword-tool exports you provide (no live data feed), never invents search volumes or rankings, never uses black-hat tactics, and writes for your country's market and spelling.
実際の使い方を見る
このAIアシスタントが、日々の業務をどう手伝えるかの例です。
会社
A UK handmade leather-goods shop
状況
The owner wants their product pages to rank for buying-intent searches
AIアシスタントの動き
- 1Reads the live product page with web_fetch, falling back to the stealth browser when the page renders client-side
- 2Hands back a paste-ready title tag and meta description with the character counts checked, in UK spelling
- 3Flags that the H1 is an image so search engines cannot read the main heading, and gives the text to add
- 4States plainly it cannot publish to the operator's CMS and notes exactly where to paste each change
- 5Refuses to invent a current ranking or search volume, and names the export that would put real numbers on it
会社
A local plumbing business in Manchester
状況
The operator asks how to rank for nearby customers searching on their phones
AIアシスタントの動き
- 1Treats it as local SEO with location and 'near me' intent rather than generic national keywords
- 2Researches what actually ranks in the local pack and clusters keywords by buying intent
- 3Builds a prioritized keyword map in a Google Sheet once the operator connects Google
- 4Drafts a content brief for a 'plumber in Manchester' service page for the operator's writer to execute, without lifting competitor text
会社
An ecommerce store chasing fast results
状況
The operator asks the agent to buy backlinks and add fake review stars
AIアシスタントの動き
- 1Declines to buy links and names the Google link-spam penalty risk, offering a white-hat earned-mention outreach draft instead
- 2Refuses to add review schema for reviews that do not exist, explaining the rich-result policy violation
- 3Reads a pasted Search Console export with exec and surfaces the striking-distance keywords one nudge from page one
- 4Recommends the legitimate next steps tied to the real data rather than promising a ranking or a timeline