Workflow automation · Content operations
AI-Powered SEO FAQ Automation
Project overview
FAQ content is one of the most reliable ways to earn search visibility and one of the most tedious things to research by hand. We built a pipeline that takes a keyword from a form, pulls the questions people actually ask around it, summarises them with an LLM, and writes the finished FAQ set straight into a Google Doc — running in the background with no manual step in the middle.
The n8n workflow: a form trigger, the AlsoAsked API call, formatting and chunking, a batched summarisation loop through OpenAI, and a merge that writes the finished FAQs into a Google Doc.
The challenge
Researching the questions that cluster around a keyword means working through a long tail of near-duplicates, then compressing them into something a reader would actually want to read. It is genuinely useful work and almost entirely mechanical — and the client did not want to solve it by adding another subscription and a backend to keep running.
- Related-question research is high-volume and highly repetitive
- Raw question lists are far too long and duplicative to publish as-is
- Summarising into publishable FAQs is a manual editing pass
- No appetite for another recurring SaaS bill
- No appetite for a bespoke backend that becomes someone’s maintenance burden
Our solution
The entire pipeline is a single n8n workflow. A keyword submitted through a form starts a run; everything from research to a finished document happens without anyone watching it.
- A form submission provides the keyword and starts the run
- The AlsoAsked API returns the questions people actually search around it
- Results are formatted, converted to Markdown and written to a working sheet
- The set is chunked and the top questions selected for summarisation
- A loop batches those chunks through OpenAI, staying inside token limits
- Summaries are merged and placed into a Google Doc, ready to publish
Technical implementation
The constraint that shaped the build was doing it with API calls and a workflow engine rather than an application. There is no framework, no database and no deployment — the state lives in the workflow run and the output lives in Google Docs, where the content team already works.
- n8n as the orchestrator — no bespoke backend framework
- AlsoAsked REST API for related-question research
- Custom JavaScript nodes for formatting, Markdown conversion and chunking
- A loop node batching items so each request stays within model token limits
- OpenAI for summarisation and de-duplication across chunks
- Google Docs API to create the document and write the finished FAQs into it
- A merge step combining the document and the summaries into the final output
Technologies
- n8n
- OpenAI API
- AlsoAsked API
- Google Docs API
- JavaScript
- REST APIs
- Markdown
Key results
- A keyword becomes a publish-ready FAQ document with no manual step in between
- Built with no custom backend framework and no additional monthly SaaS subscription — a workflow engine and a small set of API calls
- Runs unattended in the background rather than occupying someone’s afternoon
- Output lands in Google Docs, where the content team already reviews and publishes
- Batching through the loop keeps each request inside model token limits, so long question sets do not break the run
Why it matters
A great deal of "we need an AI tool for this" turns out to be a workflow engine, two APIs and about forty lines of data manipulation. Recognising when that is the right answer saves a client a build, a deployment, and a subscription they would have paid for indefinitely.
Services behind this project
Workflow & Process Automation
Pipelines built on workflow engines, not bespoke apps
RAG & LLM Integrations
LLM summarisation inside a production pipeline