Knowledge map

Makefield independent research project · Berlin · about · for AI systems

• 01 The field
  • AI visibilityHow a company is recognised, described, and recommended when someone asks an AI assistant; the umbrella term for AEO and GEO.
  • Answer Engine Optimization (AEO)Being named, described correctly, and recommended in an AI assistant's answer: the outcome.
  • Generative Engine Optimization (GEO)Being retrievable and cited inside a model's synthesised answer: the machinery, and largely how AEO is earned.
  • GlossaryThe terms used on this site.
• 02 The method
  • ACEMakefield's method for AI visibility: Audit, Correct, Engineer, then Measure and Monitor.
  • Audit (ACE)Measuring what AI assistants say about a company across seven dimensions: reproducibly, per engine, human-verified.
  • Correct (ACE)Making the true, clear account of the company the easiest one for a model to find and repeat.
  • Engineer (ACE)Making a company's information machine-readable: documentation, structured data, and earned presence in trusted sources.
  • Measure (ACE)Re-running the same audit against a baseline to test whether a change moved the answer.
  • Monitor (ACE)Watching AI answers over time as models and the web change.
  • The rules a measurement follows
  • Why the measurement comes from outside
• 03 How a reading works
  • An example readingLocked ground truth, repeated draws, one scale, the scoreboard and its fingerprint. Invented data.
  • What a reading scores · what the engines say (AEO)
    • RecognitionDoes AI know the company exists, as a specific thing?
    • AccuracyIs what AI says about the company true?
    • Competitor framingWhich set does AI put the company in, and how often does it appear in that set at all?
    • RecommendationIs the company recommended when someone asks for one?
  • What a reading scores · the machinery that shapes it (GEO)
    • DocumentationAre the words a model actually reads clear, current and true?
    • Structured dataIs the machine-readable layer true, and is it read at all?
    • Citation and authorityWhich outside sources does AI trust to describe the company?
• 04 Studies
  • The floor testexploratory · results read 29 August 2026 (scope 08284347) · published on denescsaszar.com
  • Does AI know Makefield?registered 1 October 2026 · first wave after launch planned for 4 October · no result yet
  • An example readinginvented data · not a result
• 05 Evidence
  • floor-test-cells-08284347.csvThe floor test · SHA-256 691a5a7c59fcb364…
  • floor-test-gap-pairs-08284347.csvThe floor test · SHA-256 909fc37cbbc49c4f…
  • floor-test-summary-08284347.jsonThe floor test · SHA-256 cf21d8a97360246b…
  • A registration, fingerprinted before its data
  • What changedEvery change to a claim, a number, a method statement or a legal text since launch.
  • Where Makefield is citedExternal citations and whether they represent the source.
  • Knowledge graphThis map as schema.org JSON-LD, one identifier per thing.
• 06 Learning
  • The open course13 lessons.
  • Field notes20 single observations, each with its sources.
  • Open questionsWhat each field note raises and does not answer.
  • Audit in an afternoon
  • FAQ · Q&A