Skip to main content
Auto Alpha AdvisoryAuto Alpha Advisory

AI engines, investors and regulators now read all of it

Nobody in your company has read the whole thing.

The independent read of your business, before the people who have to trust it do their own.

Book my Full Read callRun my free scan

read and signed by Matt Owen, CA(SA), ex-KPMG audit · it starts with a 15-minute call

Live engine read of autoalphaadvisory.co.za, dated 2026-09-27: nine domains scored — security 97, seo 95, geo 70, performance 89, content 83, ux & design 88, technical 99, popia 99, email security 99. Overall 88 out of 100, rated Good. GPTBot server-side rendering checks all pass: H1, main copy and JSON-LD identical for GPTBot.
our own site, scored in public — including the flagsrun the same read on your site — free →
  • founder CA(SA) — writes the code
  • own engine SEO 95/100 · public
  • background ex-KPMG audit
  • base Cape Town · serves SA
what the outside reader finds

Your business is checked before anyone calls.

Customers ask ChatGPT and Google who to use. Investors and partners read your website against your documents. Regulators read your privacy policy. Here is what a read of the same surface turns up.

5 of 8

AI crawlers behind ChatGPT, Claude and Perplexity got an empty home page from one South African site that let every crawler in. A browser saw it fine.

35 of 58

findings on one full read were the company’s own team’s work: copy, settings and code, hours to weeks. On another, 39 of 56.

Share of 432 answers from four AI models that named each accounting firm, July 2026, for the three most-named firms and the Big 4: BDO 28 percent, Forvis Mazars 20 percent, PKF 16 percent, Deloitte 14 percent, PwC 13 percent, KPMG 12 percent, EY 9 percent. From the SA AI Visibility Index, Edition 1.

Who the AI names isn’t the league table.

Three mid-tier firms were named more often than every Big 4 firm.

SA AI Visibility Index, Edition 1 · method and every question published →
what I do

Read. Fix. Prove.

01 / read

Read

Your website and Google listing, who four AI models name when your buyers ask, what the AI crawlers are actually served, and your public claims checked against their source.

what the read covers →
02 / fix

Fix

Every finding comes back labelled by who can fix it and how long it takes. I fix what’s mine to fix: the site, the copy, the build.

websites and fixes →
03 / prove

Prove

I re-run the checks after the fix, so you see a dated before-and-after, finding by finding, not just a report.

my own read, weak rows included →

I hold my own site to it: 88/100 on my engine, read 27 September 2026.

A report is copyable. A dated before-and-after is not.

why every fix ends in a re-run of the checks, finding by finding — see The Full Read →

the full read

Three doors, one method.

Which door depends on who is asking. Each starts with a 15-minute call and one finding from your own platform, free. The price is fixed, in rand, before any work starts.

  1. door 1

    Your own platform

    The Full Read · Platform Remediation
    who it’s for

    A founder, COO or listed company whose site, claims and terms an investor, a regulator or ChatGPT is about to read.

    you receive

    The read and an action plan, every finding classed by who can fix it. Then the fix, if you want it, with the checks re-run, finding by finding.

  2. door 2

    Against your competitors

    The Full Read · Competitive Position
    who it’s for

    A company with named competitors it takes seriously, asking how it compares.

    you receive

    Everything in the first door, plus three named competitors read on the same instrument the same day, and a plan to lead.

  3. door 3

    A company you’re backing

    The Full Read · Investor Due Diligence
    who it’s for

    A fund, syndicate or angel with a named candidate and a decision date.

    you receive

    What was verified, what wasn’t and what was contradicted, with the questions to ask before you sign. Never a recommendation to invest.

how each door works →

the fix it leads to

Run your own business? Start with the fix.

I find what’s stopping Google and the AI assistants sending you customers, and fix it for a fixed price, agreed before I start.

SEO & GEO fixes → AI-Ready Websites → what other studios charge →

everything else I do is on the services page → · before you spend on AI inside the business: what AI really costs a South African business →

It starts with a 15-minute call.

You tell me what matters to you right now. I come back with one thing worth knowing about your business, with its fix, at no cost. If you want the rest, you get a fixed price, in rand, in writing before anything starts.

Book my Full Read callRun my free scan

measuring your AI visibility is becoming a commodity — fixing it isn’t

sources

The accounting figures are from the SA AI Visibility Index, Edition 1 (July 2026, restated 3 September 2026): 36 buyer questions put to ChatGPT, Claude, Gemini and Perplexity through their APIs, three runs each, 432 answers. The 5 of 8 (the crawlers OpenAI, Anthropic and Perplexity name), 35 of 58 and 39 of 56 figures are from full reads in September 2026, names withheld.

OpenAI’s, Anthropic’s and Perplexity’s crawlers don’t execute JavaScript; generative engines reward statistics, cited sources and expert quotes — in Princeton’s lab test they lifted a page’s visibility by 30–40%, on pages the engine had already chosen as sources. Roughly 48% of Google searches now surface an AI Overview, where organic click-through runs at 2.36% against 3.82% without one.

“GEO can boost visibility by up to 40% in generative engine responses.” — Aggarwal et al., “GEO: Generative Engine Optimization” · Princeton · KDD 2024