Your buyers ask ChatGPT, Perplexity and Google AI before they ask you. We make your business the one those engines name.
Free. Two more questions, about a minute. No call required.
The free first look gives you a scorecard like this one: your AI visibility across 10 pillars, your competitors tested on the same questions, and the three fixes we would do first.
Our rubric started as 470 auditable factors from the 2026 AEO and GEO body of work, consolidated into 72 checks across 10 pillars. Each pillar carries a weight out of 100, so the score reflects what actually decides whether an engine names you, not a flat checklist.
Whether ChatGPT, Perplexity, Google AI Overviews and Gemini actually cite you for your buyers' questions, how often, how prominently, and how that compares with 3 competitors.
Does each key page state the answer in its first lines, in a passage an engine can lift? Question-style headings, real lists and tables, FAQ blocks, depth, freshness.
Whether GPTBot, ClaudeBot, PerplexityBot, Google-Extended and CCBot are allowed in, and whether they can actually fetch your pages past the CDN. Plus indexing, rendering without JavaScript, sitemap.
Core Web Vitals in the field (LCP, INP, CLS), load time and image formats, mobile usability, HTTPS, canonicals, redirects and duplicate control.
Valid JSON-LD on the pages that matter: Organization or LocalBusiness, Service or Product, FAQPage, HowTo, and whether the values match what is visible on the page.
Whether the engines know you as one clear entity: knowledge panel, number one for your own name, identical name-address-phone everywhere, Google Business Profile, map pack, topic clusters.
Experience, expertise, authority and trust signals the engines look for: bylined authors with bios, sourced claims, case studies and reviews, contact details, a clean security record.
The fundamentals the engines still lean on: unique titles and descriptions, one true H1, a logical heading hierarchy, clean URLs, internal links with descriptive anchors, alt text.
Referring domains and their relevance, references from the sources AI trusts (Reddit, YouTube, news, communities), linkable assets, anchor distribution, toxic-link hygiene.
Search Console and GA4 set up and watched, rank tracking with baselines, and a recurring check of whether the brand is cited in AI answers so the score means something month to month.
Every score is reproducible. Same prompts, same anchors, same weights, so a re-audit shows real movement instead of a different opinion.
We write 5 to 8 questions your buyers actually ask, including the commercial ones ("best X in [city]", "X vs Y", "is [brand] legit"), and run each one live on ChatGPT, Perplexity, Google AI Overviews and Claude, several times, from clean sessions.
Each of the 72 checks gets one observed fact: a count, a URL, a measured value. If something cannot be measured, it is marked not applicable with the reason. We never invent a number.
A robots.txt that allows GPTBot means nothing if the CDN blocks it. We fetch your key pages as GPTBot, ClaudeBot and PerplexityBot and compare them with a normal browser fetch. Cloudflare's default AI-crawler block, firewall rules and bot-management plugins all show up here, and they are the single most common reason a good site is invisible.
Each check scores 0 to 100 against fixed anchor bands. A pillar is the weighted average of its checks. The overall score is the weighted average of the pillars, graded A to F. Fixes are ranked by how much weighted ground each one is losing, so the first item on your list is always the one worth the most.
The live differentiator. Whether the engines surface or cite your brand when answering your audience's questions. If you never appear, you are invisible in the channel that now mediates discovery.
Search did not disappear. It moved into a single answer most people never scroll past. If that answer does not name you, the click never happens.
Asked to ChatGPT, not typed into Google. One or two names come back. The rest are invisible.
Perplexity reads your site, your competitors' sites and every review it can find, then decides who to cite.
AI agents are starting to browse and buy on a person's behalf. If your site is not readable, they move on.
Three things decide whether AI sends people your way. We handle all three.
Start with the flagship, add the rest as they help. Nothing here is a bundle you do not need.
We get your brand named and quoted inside AI answers, so you show up the moment someone asks about what you sell.
A new wave of AI agents will browse and buy for your customers. We make your site one they can actually use.
Fast, modern sites built AI-ready from day one and built to convert the traffic AI and search send you.
When you want to buy attention on top of earning it, we run lean Google and Meta campaigns pointed at the same goal.
The method is the same. What changes is which questions your buyers ask, and which engines they ask them in.
Trades, clinics, builders, brokers, studios. Your buyer asks "who is the best near me" and takes one of the two names that come back.
Shops and product brands. Your buyer asks for a comparison, and increasingly lets an agent do the shopping.
Firms, SaaS, manufacturers. Your buyer researches for weeks and asks the engines to shortlist.
An AI agent never looks at your screen. It reads the accessibility tree, the markup and the files you publish for machines. We test that with the same six checks Google added to Lighthouse's Agentic Browsing category in May 2026, then we go further with a live task run.
Agents act on the accessibility tree. A button without a programmatic name does not exist to them. The usual offenders: modals, drawers, chat widgets, carousels with cloned slides, mobile menus.
For a person a layout shift is a misclick. For an agent it is clicking where the button no longer is, which ends the whole task.
Lighthouse parses llms.txt as markdown: it needs an H1, real descriptions and markdown links. A bare list of URLs fails, and a malformed file scores worse than none.
WebMCP lets a site expose actions (get a quote, check availability, book) with a schema an agent can call. Proposed standard, Chrome origin trial from Chrome 149.
Forms can carry declarative annotations (tool name, description, per-field descriptions) so the browser derives a tool from them. Without them the agent guesses from labels.
A registered tool with a broken JSON Schema is worse than no tool. Agents learn not to trust the site.
Being the go-to means saying what the tools can and cannot do.
No jargon and no black box. You always know what we found, what we fixed and what it moved.
We run your brand and your competitors through real AI answers and score you on the 72-point rubric across 10 pillars.
We rebuild the content, structure and signals that make you easy to find, quote and trust. Biggest wins first.
Changes go live on a fast, AI-ready site, whether we built it or you already have one.
We track your share of AI answers month over month and keep tuning as the engines shift.
Almost everyone starts with the audit. The retainer and builds come after, if the audit says they should. Pricing is scoped to your business, not a fixed shelf price.
The best first step. A full read of how AI and agents see you, with a clear fix list.
Ongoing work to grow and hold your share of AI answers, month after month.
A new site, AI-ready from day one and built to convert the traffic it earns.
Insite Found is a small team that has spent years doing search, paid media and website builds for Canadian businesses. When buyers moved into AI answers, we moved with them. This is all we do now.
Every engagement starts with the scorecard. We do not guess at fixes and we do not sell work the audit did not ask for.
Fixes are ranked by impact and effort. The first month is the cheap, high-impact list, not a strategy deck.
No jargon in the reports. Share of answer, engines tested, pages changed, leads that came in.
Content, site, accounts and data are yours from day one. Leave any time and take it all with you.
Answer engine optimization. Making your content the answer an engine quotes, not the link it lists.
Generative engine optimization. The same idea for engines that write the answer: ChatGPT, Gemini, AI Overviews.
Out of all the answers to your category's questions, the share that name you. Tracked per engine, against competitors.
What an engine believes your business is: one name, one address, one category, everywhere. Fuzzy entities get skipped.
A plain-text file at your site root, written for AI systems, that describes your key pages. Parsed as markdown.
An emerging standard that lets a site expose actions (book, quote, buy) to AI agents directly, with a schema.
Your customers now ask ChatGPT, Perplexity and Google AI who to call. AI visibility is whether those engines know your business and name it. Here is what that means, how it is measured and what moves it.
Four engines, four different habits. Here is how each one gathers sources, what it tends to trust, and what that means for a business that wants to be named.
The next buyer might not be a person on your site. It might be an AI agent booking, comparing or buying on their behalf. Here is what it takes to be a site an agent can actually use.
Start with the free first look. Your scorecard lands in your inbox within two business days.
Three quick steps. About a minute. We come back with where you stand and what we would fix first.
We are checking now. Your scorecard goes to .
Optional. Up to three, any format.