Our Approach → AI-Native Methodology

AI Tools in Every Workflow. Human Judgement on Every Decision.

We rebuilt our workflows around AI tools in 2024. The shift did not change our strategy. It changed how fast we can execute it. Here is exactly where AI enters and where it stops.

Request a Blueprint →
See the workflows
Updated April 2026
12 min read
3x Research output vs. manual process
2024 Year we rebuilt around AI tools
0% AI content published without human edit

Faster execution on research and briefing. Human editorial oversight on everything that goes live.

AI in Our Workflow

Where AI Does the Work.

Six specific workflow areas where AI tools have materially changed how fast and how accurately we execute.

Semantic clustering used to take a full day of manual work with Ahrefs and a spreadsheet. We now run AI-assisted clustering that groups keyword opportunities by intent and topic in under two hours. More thorough, fewer missed clusters.

  • Semantic keyword grouping across 200 to 500 terms
  • Search intent classification per cluster
  • Hub architecture mapped directly from cluster output

Every piece of content starts from a structured brief. AI builds the first draft: target entities, competitor gap analysis, suggested headings, schema requirements, and word count benchmarks. The brief gets reviewed by a strategist before any writing starts.

  • Entity targets and semantic coverage requirements
  • Top competitor content compared and gaps flagged
  • Schema type and FAQ structure recommendations

Raw data from GA4, GSC, and Ahrefs tells you what happened. AI synthesis tells you why and where to focus next. We run AI-assisted analysis to turn raw metric exports into prioritised action lists rather than lengthy reports that nobody reads.

  • GSC and GA4 data synthesised into insight, not tables
  • Ranking movement patterns analysed across clusters
  • Opportunity identification based on current gaps

AI answer engines like Perplexity, ChatGPT, and Google AI Overviews assess topical authority and citation-worthiness before surfacing a source. We structure pages to answer specific questions directly, use schema to signal structure, and test responses across answer engines before and after optimisation.

  • Question-answer structure mapped for each page
  • FAQ and HowTo schema built in systematically
  • Perplexity and ChatGPT citation testing built into QBRs

GEO goes a step beyond AEO. The goal is to appear as a recognised entity in knowledge graphs and AI training data, not just to answer individual questions well. We map brand entity relationships, build structured citations, and ensure consistent brand signals across the web.

  • Brand entity recognition audit (Google Knowledge Panel)
  • Citation consistency across directories and publications
  • Structured data entity linking across page types

Writing schema markup by hand for dozens of pages is error-prone and slow. We use AI-assisted schema generation to produce valid JSON-LD across all page types: WebPage, FAQPage, HowTo, Article, BreadcrumbList, and Organization. Every schema block is validated before deployment.

  • Full JSON-LD schema generated for every new page
  • Schema type selected based on page intent
  • Validated against Google Rich Results Test
SEO vs. AEO vs. GEO

Three Search Channels. All Three Matter Now.

When we started in 2020, search meant Google rankings. You optimised for position one and measured click-through rate. That model still matters, but it no longer covers the full picture.

AI-generated answers now appear above the blue links for many commercial queries. Answer engines like Perplexity handle a growing share of research-intent searches. Traditional SEO does not prepare you for either of these channels.

We optimise for all three simultaneously. The same hub architecture that builds Google authority also creates the topical depth that answer engines look for. The entity work that supports GEO reinforces the schema signals that support AEO.

Traditional SEO
Position tracking
Click-through rate
Backlink acquisition
Page speed / CWV
AEO
Question-answer structure
FAQ / HowTo schema
AI citation testing
Topical authority depth
GEO
Entity recognition in knowledge graph
Consistent brand signals across web
Structured entity linking in markup
AI training data citation presence
The Line Between AI and Human

What AI Does. What We Do.

The agencies that use AI badly are the ones that let it do the thinking as well as the doing. Research and briefing are automatable. Commercial judgement is not.

Deciding whether to compete for a keyword cluster your site cannot currently win. Knowing when to recommend paid support because organic alone will not be fast enough. Spotting that a client's conversion problem is actually a positioning problem. None of that comes from a prompt.

We are transparent with clients about where AI enters our workflow. If you want to know which tools we use and at which stage, ask us in the discovery call. We will walk you through it.

Common Questions

Questions About AI in Our Work.

What is AEO and how does it differ from SEO?

AEO stands for Answer Engine Optimisation. Where traditional SEO targets ranking positions in Google's blue-link results, AEO targets citations and featured placements in AI-generated answers from tools like ChatGPT, Perplexity, Google AI Overviews, and Gemini. AEO involves structuring content to directly answer specific questions, using schema markup extensively, and building the topical authority that language models recognise as a reliable source worth citing.

What is GEO in digital marketing?

GEO stands for Generative Engine Optimisation. It is the practice of optimising content so that AI systems which generate answers are more likely to surface your brand or content in their responses. GEO builds on AEO with additional focus on entity recognition, structured data, and ensuring your brand appears as a clear entity in knowledge graphs and AI training data.

Does Viaduct Generation use AI to write client content?

We use AI to accelerate research, create structured briefs, and identify content gaps. Final content is written and edited by human strategists and editors. AI-generated content that has not been substantially reviewed and developed by a human does not meet our quality standards and does not perform consistently in either traditional search or AI answer engines. We are an AI-native agency, not an AI content factory.

Which AI tools does Viaduct Generation use?

Our core stack includes Claude and GPT-4 for research synthesis and brief production, Perplexity for answer engine testing, Semrush and Ahrefs with AI features for keyword research and competitor analysis, and custom workflows built around GA4 and Google Search Console data. We also use AI-assisted schema generation and entity mapping tools. The stack evolves, and we assess new tools against one question: does this make our work faster without making it worse?

Our Approach Hub

More From Our Approach.

Pillar Page
Our Approach Overview

The full picture: sprint model, AI tools, attribution, and why we keep the client roster small.

View pillar →
Spoke 01
The 90-Day Sprint Model

Six phases, fixed scope, and a QBR at close of every cycle.

Read more →
Spoke 03
Revenue Attribution: How We Model It

Multi-touch attribution, CRM integration, and making organic search accountable.

Read more →
Spoke 04
Why We Only Work with Ambitious Businesses

What makes a client a good fit and why selectivity produces better results for everyone.

Read more →
AI-Native From Day One

See the Methodology Applied.

A Search Growth Blueprint covers your technical health, content gaps, AEO opportunities, and a 90-day sprint plan. Free for businesses that are a good fit.

Request Your Blueprint →
See client results