Here’s how the Ahrefs international marketing team uses Agent A to automate their work.

International marketing is a job that gets exponentially complicated with each additional region and language you add.

For example, we publish the Ahrefs blog in eight languages, which means roughly every meaningful task—refreshing an old article, checking hreflang tags, swapping internal links, tracking what a competitor just shipped—has to be done eight times, or eight times a month, or eight times per product launch. Thankfully, we have some incredible marketers on the case.

Erik Sarissky is Head of International Marketing & Product Localization at Ahrefs. He runs the international marketing team at Ahrefs, leading a small-but-mighty cadre of international marketers responsible for scaling growth in regions like Spain, France, Germany, and Japan.

Takanori Kawaharada is Ahrefs’ Regional Head of Marketing, Japan. Taka is responsible for all of Ahrefs’ marketing channels in Japan: events and webinars, content creation, social media, product marketing, and much more besides.

These are some of the AI tools Erik and Taka (and the rest of the international marketing team) have built with our marketing agent, Agent A. Each explanation includes a starter prompt you can paste into a fresh workspace to build your own version.

What is Agent A?

Agent A is a marketing agent from Ahrefs—an AI assistant with direct access to the full Ahrefs dataset that can carry out marketing tasks autonomously, rather than just answer questions.

Agent A includes:

  • Unrestricted access to Ahrefs endpoints. Every endpoint we use to build Ahrefs is available, including many you cannot reach via API or MCP.
  • Serious tech stack underneath. Postgres for state, Flask for UIs, an OpenRouter proxy with 300+ models, web fetch with full-page parsing, PDFs, OCR, scheduled jobs.
  • Native connectors to marketing tools. Slack, HubSpot, GitHub, Notion, Linear, Mailchimp, Resend, SendGrid, Stripe, Gong, WordPress, Airtable, Apify, and even Semrush.
  • Expert skill library. The Ahrefs team has contributed pre-built marketing skills and applications that encode how we actually work.
Hreflang is a very dull, very important part of international SEO. This tool generates hreflang tag sets for any international site:

For any canonical URL, the tool returns the complete set of hreflang tags (<link rel="alternate" hreflang="..." href="...">), tailored to the international markets you care about, and ready to paste into a CMS field or a sitemap generator.

Audit mode also allows you to troubleshoot common hreflang issues. Point it at a page that already has hreflang tags and it tells you what’s wrong—missing reciprocal tags, wrong language codes, broken alternates, missing x-default. This catches the half-implemented-by-the-previous-agency state most international sites are in.

Starter prompt

Build me a hreflang generator + auditor for international sites. Support both URL patterns: path-based (ahrefs.com/blog/de/slug) and subdomain-based (de.example.com/blog/slug) via a single slug-prefix parser. Slug matching in order: (1) exact same-slug across languages, (2) Ahrefs Site Audit hreflang_audit endpoint when available (cache to disk), (3) title-similarity fallback. For a canonical URL, output the full <link rel=“alternate” hreflang> tag block including x-default. Audit mode: point at a page that already has hreflang and report missing reciprocals, wrong language codes, broken alternates. Auto-import brand configurations from my Intl Blog Monitor so I don’t reconfigure the same site twice.

6. Run real-time English ↔ Japanese subtitles at conferences

Twice a year Taka hosts or sponsors an event where the speaker is in one language and half the audience is in the other. Professional simultaneous interpreters cost $2,000+/day and don’t know our product. The Live Interpreter is what Taka opens on a laptop next to the stage:

This translation tool records live audio and generates subtitles on-the-fly using Gemini 3 Flash (with a choice of other models for different situations: Claude Haiku is more accurate but too slow for live events).

At Taka’s specification, the subtitling tool forces a grammatically complete output sentence even when the input is a mid-phrase chunk of live speech: no noun-ending sentences, no trailing particles, and must close with a proper Japanese predicate (the polite-form copula or a verb conjugation).

And crucially from a product perspective: a dedicated SEO/Ahrefs glossary is injected at the top of the system prompt with the literal directive “ABSOLUTE HIGHEST PRIORITY — overrides everything else”. So “AI Overviews” always renders as the one approved Japanese term we use internally, never a near-synonym. “Backlink”, “Domain Rating”, and every other brand-controlled term map to exactly one Japanese rendering.

Starter prompt

Build me a real-time conference interpreter for [my two languages]. Browser captures audio chunks, POSTs each transcribed chunk to /translate. Server detects direction by character-set / language ID. System prompt enforces two non-negotiables: (1) every output sentence must be grammatically complete with a proper predicate even when the input is a mid-phrase chunk — if input cuts off, model adds a closing verb / clause to finish the sentence so subtitles never dangle; (2) inject my domain glossary (both directions) at the top of the system prompt with the literal directive “ABSOLUTE HIGHEST PRIORITY — overrides everything else” so brand terms never get substituted with synonyms. Pre-wire 4 model options ordered by latency, not accuracy: Gemini Flash Lite, Gemini 3 Flash (default), Claude Haiku 4.5, GPT-4.1 Nano. max_tokens=400, temperature=0.2. Show direction + model in the UI; switch model on the fly without reloading.

Final thoughts

Ahrefs creates a ton of marketing collateral each month, and the international marketing team has a mountain to climb to localize all of our blog posts, emails, product videos, YouTube tutorials, and run live events.

Erik and Taka use Agent A to automate big chunks of the less-skilled, more repetitive work each month. All of these tools started life as manual, time-consuming workflows and turned into an app over a chat session or two. The total work was probably a normal week’s worth of an engineer’s time, except that neither Erik nor Taka is an engineer.

If you’re an Ahrefs customer, Agent A is free to try for one month. Paste any of the starter prompts above into a fresh workspace and your Agent A will build the tool—with your data, your locales, and your priorities.

Further reading

Similar Posts